From 7d06b137b19f82df5f0850a14def214676356476 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Fri, 27 Jun 2025 07:34:59 -0700
Subject: [PATCH 01/23] First pass updates for transformer engine support

---
 .../models/transolver/Physics_Attention.py    | 346 ++++++++++++++++++
 physicsnemo/models/transolver/transolver.py   | 103 ++++--
 2 files changed, 414 insertions(+), 35 deletions(-)

diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index 519935e900..c4b7807d7c 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -31,9 +31,355 @@
 
 import torch
 import torch.nn as nn
+import transformer_engine.pytorch as te  # noqa: F401
 from einops import rearrange
 
 
+class Physics_Attention_Base(nn.Module):
+    """
+    Base class for all physics attention modules.
+
+    Implements key functionality that is common across domains:
+    - Slice weighting and computation
+    - Attention among slices
+    - Deslicing
+    - Output Projection
+
+    Each subclass must implement it's own methods for projecting input domain tokens onto the slice space.
+
+    Deliberately, there are not default values for any of the parameters.  It's assumed you will
+    assign them in the subclass.
+
+    """
+
+    def __init__(self, dim, heads, dim_head, dropout, slice_num):
+        super().__init__()
+        inner_dim = dim_head * heads
+        self.dim_head = dim_head
+        self.heads = heads
+
+        self.scale = dim_head**-0.5
+
+        self.softmax = nn.Softmax(dim=-1)
+        self.dropout = nn.Dropout(dropout)
+        self.temperature = nn.Parameter(torch.ones([1, heads, 1, 1]) * 0.5)
+
+        self.in_project_slice = nn.Linear(dim_head, slice_num)
+        for l_i in [self.in_project_slice]:
+            torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
+
+        self.to_q = nn.Linear(dim_head, dim_head, bias=False)
+        self.to_k = nn.Linear(dim_head, dim_head, bias=False)
+        self.to_v = nn.Linear(dim_head, dim_head, bias=False)
+
+        # # These are used in the transformer engine pass function:
+        # self.qkv_project = nn.Linear(dim_head, 3 * dim_head, bias=False)
+        # self.attn_fn = te.DotProductAttention(num_attention_heads=self.heads,
+        #                                       kv_channels= self.dim_head,
+        #                                       attention_dropout=dropout,
+        #                                       qkv_format="bshd",
+        #                                       softmax_scale=self.scale)
+
+        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
+
+    def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
+        """
+        Project the input onto the slice space.
+        """
+        raise NotImplementedError("Subclasses must implement this method")
+
+    def compute_slices_from_projections(
+        self, slice_projections: torch.Tensor, fx: torch.Tensor
+    ) -> tuple[torch.Tensor, torch.Tensor]:
+        """
+        Compute slice weights and slice tokens from input projections and latent features.
+
+        Args:
+            slice_projections (torch.Tensor):
+                The projected input tensor of shape [Batch, N_heads, N_tokens, Slice_num],
+                representing the projection of each token onto each slice for each attention head.
+            fx (torch.Tensor):
+                The latent feature tensor of shape [Batch, N_heads, N_tokens, Head_dim],
+                representing the learned states to be aggregated by the slice weights.
+
+        Returns:
+            tuple[torch.Tensor, torch.Tensor]:
+                - slice_weights: Tensor of shape [Batch, N_heads, N_tokens, Slice_num],
+                representing the normalized weights for each slice per token and head.
+                - slice_token: Tensor of shape [Batch, N_heads, Slice_num, Head_dim],
+                representing the aggregated latent features for each slice, head, and batch.
+
+        Notes:
+            - The function first computes a temperature-scaled softmax over the slice projections to obtain slice weights.
+            - It then aggregates the latent features (fx) for each slice using these weights.
+            - The aggregated features are normalized by the sum of weights for numerical stability.
+        """
+        # Project the latent space vectors on to the weight computation space,
+        # and compute a temperature adjusted softmax.
+        slice_weights = nn.functional.softmax(
+            slice_projections / torch.clamp(self.temperature, min=0.1, max=5), dim=-1
+        )  # [Batch, N_heads, N_tokens, Slice_num]
+
+        # Average the slices over the token dimension
+        slice_norm = slice_weights.sum(2)  # [Batch, N_heads, Slice_num]
+
+        # This does the projection of the latent space fx by the weights:
+        slice_token = torch.matmul(slice_weights.transpose(2, 3), fx)
+
+        # Apply the normalization (summed weights)
+        slice_token = slice_token / ((slice_norm[:, :, :, None] + 1e-5))  # B H G D
+
+        return slice_weights, slice_token
+
+    def compute_slice_attention(self, slice_tokens: torch.Tensor) -> torch.Tensor:
+        """
+        Compute an attention mechansism for the slices
+        """
+        q_slice_token = self.to_q(slice_tokens)
+        k_slice_token = self.to_k(slice_tokens)
+        v_slice_token = self.to_v(slice_tokens)
+
+        dots = torch.matmul(q_slice_token, k_slice_token.transpose(-1, -2)) * self.scale
+        attn = self.softmax(dots)
+        attn = self.dropout(attn)
+        out_slice_token = torch.matmul(attn, v_slice_token)  # B H G D
+
+        return out_slice_token
+
+    def compute_slice_attention2(self, slice_tokens: torch.Tensor) -> torch.Tensor:
+        """
+        TE implementation of slice attention
+        """
+
+        qkv = self.qkv_project(slice_tokens)
+        qkv = rearrange(qkv, " b h s (t d) -> t b s h d", t=3, d=self.dim_head)
+        q_slice_token, k_slice_token, v_slice_token = qkv.unbind(0)
+
+        out_slice_token2 = self.attn_fn(q_slice_token, k_slice_token, v_slice_token)
+        out_slice_token2 = rearrange(
+            out_slice_token2, "b s (h d) -> b h s d", h=self.heads, d=self.dim_head
+        )
+
+        return out_slice_token2
+
+    def compute_slice_attention3(self, slice_tokens: torch.Tensor) -> torch.Tensor:
+        """
+        Torch SDPA implementation of slice attention
+        """
+
+        # qkv = self.qkv_project(slice_tokens)
+        # qkv = rearrange(qkv, " b h s (t d) -> t b h s d", t=3, d=self.dim_head)
+        q_slice_token = self.to_q(slice_tokens)
+        k_slice_token = self.to_k(slice_tokens)
+        v_slice_token = self.to_v(slice_tokens)
+        # q_slice_token, k_slice_token, v_slice_token = qkv.unbind(0)
+
+        out_slice_token3 = torch.nn.functional.scaled_dot_product_attention(
+            q_slice_token, k_slice_token, v_slice_token, is_causal=True
+        )
+
+        return out_slice_token3
+
+    def project_attention_outputs(
+        self, out_slice_token: torch.Tensor, slice_weights: torch.Tensor
+    ) -> torch.Tensor:
+        """
+        Project the attended slice tokens back onto the original token space.
+
+        Args:
+            out_slice_token (torch.Tensor):
+                The output tensor from the attention mechanism over slices,
+                of shape [Batch, N_heads, Slice_num, Head_dim].
+            slice_weights (torch.Tensor):
+                The slice weights tensor of shape [Batch, N_heads, N_tokens, Slice_num],
+                representing the contribution of each slice to each token.
+
+        Returns:
+            torch.Tensor:
+                The reconstructed output tensor of shape [Batch, N_tokens, N_heads * Head_dim],
+                representing the attended features for each token, with all heads concatenated.
+
+        Notes:
+            - The function projects the attended slice tokens back to the token space using the slice weights.
+            - The output is reshaped to concatenate all attention heads for each token.
+        """
+
+        out_x = torch.matmul(slice_weights, out_slice_token)
+        out_x = rearrange(out_x, "b h n d -> b n (h d)")
+        return self.to_out(out_x)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        """
+        Forward pass of the Physics Attention module.
+
+        Input x should have shape of [Batch, N_tokens, N_Channels] ([B, N, C])
+        """
+
+        # Project the inputs onto learned spaces:
+        x_mid, fx_mid = self.project_input_onto_slices(x)
+        # x_mid and fx_mid should have shapes of [B, N_head, N_tokens, Head_dim]
+
+        # Perform the linear projection of learned latent space onto slices:
+        slice_projections = self.in_project_slice(x_mid)
+
+        # Slice projections has shape [B, N_head, N_tokens, Head_dim], but head_dim may have changed!
+
+        # Use the slice projections and learned spaces to compute the slices, and their weights:
+        slice_weights, slice_tokens = self.compute_slices_from_projections(
+            slice_projections, fx_mid
+        )
+
+        # slice_weights has shape [Batch, N_heads, N_tokens, Slice_num]
+        # slice_tokens has shape  [Batch, N_heads, N_tokens, head_dim]
+
+        # Apply attention to the slice tokens
+        # out_slice_token = self.compute_slice_attention(slice_tokens)
+        # out_slice_token = self.compute_slice_attention2(slice_tokens)
+        out_slice_token = self.compute_slice_attention3(slice_tokens)
+        # Shape unchanged
+
+        # Deslice:
+        outputs = self.project_attention_outputs(out_slice_token, slice_weights)
+
+        # Outputs now has the same shape as the original input x
+
+        return outputs
+
+
+class Physics_Attention_Irregular_Mesh_2(Physics_Attention_Base):
+    """
+    Specialization of PhysicsAttention to Irregular Meshes
+    """
+
+    def __init__(self, dim, heads=8, dim_head=64, dropout=0.0, slice_num=64):
+        super().__init__(dim, heads, dim_head, dropout, slice_num)
+        inner_dim = dim_head * heads
+
+        self.in_project_x = nn.Linear(dim, inner_dim)
+        self.in_project_fx = nn.Linear(dim, inner_dim)
+
+    def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
+        """
+        Project the input onto the slice space.
+        """
+
+        fx_mid = rearrange(
+            self.in_project_fx(x), "B N (h d) -> B h N d", h=self.heads, d=self.dim_head
+        )
+        x_mid = rearrange(
+            self.in_project_x(x), "B N (h d) -> B h N d", h=self.heads, d=self.dim_head
+        )
+
+        return x_mid, fx_mid
+
+
+class Physics_Attention_Structured_Mesh_2D_2(Physics_Attention_Base):
+    """
+    Specialization for 2d image-like meshes
+    """
+
+    def __init__(
+        self,
+        dim,
+        heads=8,
+        dim_head=64,
+        dropout=0.0,
+        slice_num=64,
+        H=101,
+        W=31,
+        kernel=3,
+    ):  # kernel=3):
+        super().__init__(dim, heads, dim_head, dropout, slice_num)
+
+        inner_dim = dim_head * heads
+        self.H = H
+        self.W = W
+
+        self.in_project_x = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
+        self.in_project_fx = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
+
+    def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
+
+        # Rearrange the input tokens back to an image shape:
+        x = rearrange(x, "b (h w) c -> b c h w", h=self.H, w=self.W)
+
+        # Apply the projections, here they are convolutions in 2D:
+        input_projected_fx = self.in_project_fx(x)
+        input_projected_x = self.in_project_x(x)
+
+        # Next, re-reshape the projections into token-like shapes:
+        input_projected_fx = rearrange(
+            input_projected_fx,
+            "b (n_heads head_dim) h w -> b n_heads (h w) head_dim",
+            head_dim=self.dim_head,
+            n_heads=self.heads,
+        )
+        input_projected_x = rearrange(
+            input_projected_x,
+            "b (n_heads head_dim) h w -> b n_heads (h w) head_dim",
+            head_dim=self.dim_head,
+            n_heads=self.heads,
+        )
+
+        # Return the projections:
+        return input_projected_x, input_projected_fx
+
+
+class Physics_Attention_Structured_Mesh_3D_2(Physics_Attention_Base):
+    """
+    Specialization for 3D-image like meshes
+    """
+
+    def __init__(
+        self,
+        dim,
+        heads=8,
+        dim_head=64,
+        dropout=0.0,
+        slice_num=32,
+        H=32,
+        W=32,
+        D=32,
+        kernel=3,
+    ):
+        super().__init__(dim, heads, dim_head, dropout, slice_num)
+
+        inner_dim = dim_head * heads
+        self.H = H
+        self.W = W
+        self.D = D
+
+        self.in_project_x = nn.Conv3d(dim, inner_dim, kernel, 1, kernel // 2)
+        self.in_project_fx = nn.Conv3d(dim, inner_dim, kernel, 1, kernel // 2)
+
+    def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
+        """
+        Project the input onto the slice space.
+        """
+
+        x = rearrange(x, "b (h w) c -> b c h w", h=self.H, w=self.W)
+
+        # Apply the projections, here they are convolutions:
+        input_projected_fx = self.in_project_fx(x)
+        input_projected_x = self.in_project_x(x)
+
+        # Next, re-reshape the projections into token-like shapes:
+        input_projected_fx = rearrange(
+            input_projected_fx,
+            "b (n_heads head_dim) h w -> b n_heads (h w) head_dim",
+            head_dim=self.dim_head,
+            n_heads=self.heads,
+        )
+        input_projected_x = rearrange(
+            input_projected_x,
+            "b (n_heads head_dim) h w -> b n_heads (h w) head_dim",
+            head_dim=self.dim_head,
+            n_heads=self.heads,
+        )
+
+        return input_projected_x, input_projected_fx
+
+
 class Physics_Attention_Irregular_Mesh(nn.Module):
     "for irregular meshes in 1D, 2D or 3D space"
 
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index d722bc0ced..875c31b350 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -34,14 +34,18 @@
 import numpy as np
 import torch
 import torch.nn as nn
-from timm.layers import trunc_normal_
+import transformer_engine.pytorch as te
 
 import physicsnemo  # noqa: F401 for docs
 
 from ..meta import ModelMetaData
 from ..module import Module
 from .Embedding import timestep_embedding
-from .Physics_Attention import Physics_Attention_Structured_Mesh_2D
+
+# from .Physics_Attention import Physics_Attention_Structured_Mesh_2D
+from .Physics_Attention import (
+    Physics_Attention_Structured_Mesh_2D_2 as Physics_Attention_Structured_Mesh_2D,
+)
 
 ACTIVATION = {
     "gelu": nn.GELU,
@@ -107,7 +111,7 @@ def __init__(
     ):
         super().__init__()
         self.last_layer = last_layer
-        self.ln_1 = nn.LayerNorm(hidden_dim)
+        self.ln_1 = te.LayerNorm(hidden_dim)
         self.Attn = Physics_Attention_Structured_Mesh_2D(
             hidden_dim,
             heads=num_heads,
@@ -118,24 +122,34 @@ def __init__(
             W=W,
         )
 
-        self.ln_2 = nn.LayerNorm(hidden_dim)
-        self.mlp = MLP(
-            hidden_dim,
-            hidden_dim * mlp_ratio,
-            hidden_dim,
-            n_layers=0,
-            res=False,
-            act=act,
+        self.ln_mlp1 = te.LayerNormMLP(
+            hidden_size=hidden_dim,
+            ffn_hidden_size=hidden_dim * mlp_ratio,
         )
+
+        # self.ln_2 = te.LayerNorm(hidden_dim)
+        # self.mlp = MLP(
+        #     hidden_dim,
+        #     hidden_dim * mlp_ratio,
+        #     hidden_dim,
+        #     n_layers=0,
+        #     res=False,
+        #     act=act,
+        # )
         if self.last_layer:
-            self.ln_3 = nn.LayerNorm(hidden_dim)
-            self.mlp2 = nn.Linear(hidden_dim, out_dim)
+            self.ln_mlp2 = te.LayerNormLinear(
+                in_features=hidden_dim, out_features=out_dim
+            )
+            # self.ln_3 = te.LayerNorm(hidden_dim)
+            # self.mlp2 = nn.Linear(hidden_dim, out_dim)
 
     def forward(self, fx):
         fx = self.Attn(self.ln_1(fx)) + fx
-        fx = self.mlp(self.ln_2(fx)) + fx
+        fx = self.ln_mlp1(fx) + fx
+        # fx = self.mlp(self.ln_2(fx)) + fx
         if self.last_layer:
-            return self.mlp2(self.ln_3(fx))
+            return self.ln_mlp2(fx)
+            # return self.mlp2(self.ln_3(fx))
         else:
             return fx
 
@@ -192,7 +206,6 @@ def __init__(
             self.time_fc = nn.Sequential(
                 nn.Linear(n_hidden, n_hidden), nn.SiLU(), nn.Linear(n_hidden, n_hidden)
             )
-
         self.blocks = nn.ModuleList(
             [
                 Transolver_block(
@@ -220,10 +233,10 @@ def initialize_weights(self):
 
     def _init_weights(self, m):
         if isinstance(m, nn.Linear):
-            trunc_normal_(m.weight, std=0.02)
+            nn.init.trunc_normal_(m.weight, std=0.02)
             if isinstance(m, nn.Linear) and m.bias is not None:
                 nn.init.constant_(m.bias, 0)
-        elif isinstance(m, (nn.LayerNorm, nn.BatchNorm1d)):
+        elif isinstance(m, (nn.LayerNorm, te.LayerNorm, nn.BatchNorm1d)):
             nn.init.constant_(m.bias, 0)
             nn.init.constant_(m.weight, 1.0)
 
@@ -255,26 +268,43 @@ def get_grid(self, batchsize=1):
         return pos
 
     def forward(self, x, fx, T=None):
-        if self.unified_pos:
-            x = (
-                self.pos.repeat(x.shape[0], 1, 1, 1)
-                .reshape(x.shape[0], self.H * self.W, self.ref * self.ref)
-                .to(x.device)
-            )
-        if fx is not None:
-            fx = torch.cat((x, fx), -1)
-            fx = self.preprocess(fx)
-        else:
-            fx = self.preprocess(x)
-            fx = fx + self.placeholder[None, None, :]
+        with torch.autograd.profiler.record_function("prepeocess"):
+            if self.unified_pos:
+                # print(f"Applying unified pos, x shape: {x.shape}")
+                if self.pos.device != x.device:
+                    # move the position tensor:
+                    self.pos = self.pos.to(x.device)
+
+                x = (
+                    self.pos.repeat(x.shape[0], 1, 1, 1)
+                    .reshape(x.shape[0], self.H * self.W, self.ref * self.ref)
+                    .to(x.device)
+                )
+                # print(f"self.ref is {self.ref}")
+                # print(f"x shape after unified pos: {x.shape}")
+            if fx is not None:
+                # print(f"Applying fx, fx shape: {fx.shape}")
+                fx = torch.cat((x, fx), -1)
+                # print(f"fx shape after cat: {fx.shape}")
+                fx = self.preprocess(fx)
+                # print(f"fx shape after preprocess: {fx.shape}")
+            else:
+                # print(f"Applying just x, x shape: {x.shape}")
+                fx = self.preprocess(x)
+                # print(f"fx shape after preprocess: {fx.shape}")
+                fx = fx + self.placeholder[None, None, :]
+                # print(f"fx shape after placeholder: {fx.shape}")
 
         if T is not None:
-            Time_emb = timestep_embedding(T, self.n_hidden).repeat(1, x.shape[1], 1)
-            Time_emb = self.time_fc(Time_emb)
-            fx = fx + Time_emb
+            # print(f"Applying T, T shape: {T.shape}")
+            with torch.autograd.profiler.record_function("Time_Embedding"):
+                Time_emb = timestep_embedding(T, self.n_hidden).repeat(1, x.shape[1], 1)
+                Time_emb = self.time_fc(Time_emb)
+                fx = fx + Time_emb
 
-        for block in self.blocks:
-            fx = block(fx)
+        for i, block in enumerate(self.blocks):
+            with torch.autograd.profiler.record_function(f"Block_{i}"):
+                fx = block(fx)
 
         return fx
 
@@ -396,6 +426,9 @@ def forward(
             The output tensor.
 
         """
+        # print(f"Input x shape: {x.shape}")
+        # print(f"Input fx shape: {fx.shape if fx is not None else None}")
+        # print(f"Input T shape: {T.shape if T is not None else None}")
         y = self.model(x, fx, T)
         y = y.reshape(x.shape[0], self.H, self.W, -1)
         return y

From bb770a35b3b59e1932e20f32e50be4bf192bff21 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Mon, 30 Jun 2025 13:06:25 -0700
Subject: [PATCH 02/23] Update Transolver to optionally enable transformer
 engine, and make the 'fix' version of the example easier to run, scalable,
 usable at full resolution, and faster

---
 examples/cfd/darcy_transolver/config_fix.yaml |  26 +-
 .../darcy_transolver/darcy_datapipe_fix.py    | 137 ++---
 .../train_transolver_darcy_fix.py             | 484 +++++++++++-------
 .../cfd/darcy_transolver/validator_fix.py     |  73 ++-
 .../models/transolver/Physics_Attention.py    |  61 ++-
 physicsnemo/models/transolver/transolver.py   |  89 ++--
 6 files changed, 544 insertions(+), 326 deletions(-)

diff --git a/examples/cfd/darcy_transolver/config_fix.yaml b/examples/cfd/darcy_transolver/config_fix.yaml
index dd97eeb8d7..7ac13ce133 100644
--- a/examples/cfd/darcy_transolver/config_fix.yaml
+++ b/examples/cfd/darcy_transolver/config_fix.yaml
@@ -20,6 +20,15 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
+output_dir: ./output/darcy_transolver_fix
+run_id: fp32_dev_test_fullres
+
+data:
+  train_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth1.npz
+  test_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth2.npz
+  resolution: 421 #421, 211, 141, 106, 85 all viable
+  batch_size: 8 # This is the GLOBAL batch size
+
 model:
   space_dim: 2
   n_layers: 8
@@ -31,12 +40,13 @@ model:
   mlp_ratio: 1
   fun_dim: 1
   out_dim: 1
-  slice_dim: 32
+  # slice_dim: 64
   ref: 8
-  unified_pos: 1
-  slice_num: 64
+  unified_pos: False
+  slice_num: 256
+  transformer_engine: False
   
-
+precision: fp32
 
 normaliser:
   permeability:
@@ -53,12 +63,10 @@ scheduler:
   decay_pseudo_epochs: 8
 
 training:
-  resolution: 85
-  batch_size: 4
   rec_results_freq : 100
-  max_pseudo_epochs: 500
-  pseudo_epoch_sample_size: 1000
+  max_pseudo_epochs: 200
+  pseudo_epoch_sample_size: 1024
 
 validation:
   sample_size: 200
-  validation_pseudo_epochs: 2
+  validation_pseudo_epochs: 1
diff --git a/examples/cfd/darcy_transolver/darcy_datapipe_fix.py b/examples/cfd/darcy_transolver/darcy_datapipe_fix.py
index be13843946..530248378d 100644
--- a/examples/cfd/darcy_transolver/darcy_datapipe_fix.py
+++ b/examples/cfd/darcy_transolver/darcy_datapipe_fix.py
@@ -14,33 +14,20 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import sys
 from dataclasses import dataclass
 from typing import Dict, Tuple, Union
 
 import numpy as np
 import torch
-import warp as wp
 import scipy.io as scio
+import torch.distributed as dist
 
 from physicsnemo.datapipes.datapipe import Datapipe
 from physicsnemo.datapipes.meta import DatapipeMetaData
-from physicsnemo.datapipes.benchmarks.kernels.finite_difference import (
-    darcy_mgrid_jacobi_iterative_batched_2d,
-    mgrid_inf_residual_batched_2d,
-)
-from physicsnemo.datapipes.benchmarks.kernels.initialization import (
-    init_uniform_random_4d,
-)
-from physicsnemo.datapipes.benchmarks.kernels.utils import (
-    bilinear_upsample_batched_2d,
-    fourier_to_array_batched_2d,
-    threshold_3d,
-)
 
 Tensor = torch.Tensor
-# TODO unsure if better to remove this. Keeping this in for now
-wp.init()
+
+from physicsnemo.utils.profiling import profile
 
 
 class UnitTransformer:
@@ -138,6 +125,7 @@ class Darcy2D_fix(Datapipe):
         Incompatable multi-grid and resolution settings
     """
 
+    @profile
     def __init__(
         self,
         resolution: int = 256,
@@ -148,13 +136,14 @@ def __init__(
         max_iterations: int = 30000,
         convergence_threshold: float = 1e-6,
         iterations_per_convergence_check: int = 1000,
-        nr_multigrids: int = 4,
-        normaliser: Union[Dict[str, Tuple[float, float]], None] = None,
+        # nr_multigrids: int = 4,
+        # normaliser: Union[Dict[str, Tuple[float, float]], None] = None,
         device: Union[str, torch.device] = "cuda",
         train_path: str = None,
         is_test: bool = False,
         x_normalizer: UnitTransformer = None,
         y_normalizer: UnitTransformer = None,
+        downsample: int = 5,
     ):
         super().__init__(meta=MetaData())
 
@@ -167,15 +156,15 @@ def __init__(
         self.max_iterations = max_iterations
         self.convergence_threshold = convergence_threshold
         self.iterations_per_convergence_check = iterations_per_convergence_check
-        self.nr_multigrids = nr_multigrids
-        self.normaliser = normaliser
+        # self.nr_multigrids = nr_multigrids
+        # self.normaliser = normaliser
 
-        # check normaliser keys
-        if self.normaliser is not None:
-            if not {"permeability", "darcy"}.issubset(set(self.normaliser.keys())):
-                raise ValueError(
-                    "normaliser need to have keys permeability and darcy with mean and std"
-                )
+        # # check normaliser keys
+        # if self.normaliser is not None:
+        #     if not {"permeability", "darcy"}.issubset(set(self.normaliser.keys())):
+        #         raise ValueError(
+        #             "normaliser need to have keys permeability and darcy with mean and std"
+        #         )
 
         # Set up device for warp, warp has same naming convention as torch.
         if isinstance(device, torch.device):
@@ -185,31 +174,22 @@ def __init__(
         # spatial dims
         self.dx = 1.0 / (self.resolution + 1)  # pad edges by 1 for multi-grid
         self.dim = (self.batch_size, self.resolution + 1, self.resolution + 1)
-        self.fourier_dim = (
-            4,
-            self.batch_size,
-            self.nr_permeability_freq,
-            self.nr_permeability_freq,
-        )
 
         # assert resolution is compatible with multi-grid method
         # if (resolution % 2 ** (nr_multigrids - 1)) != 0:
         #     raise ValueError("Resolution is incompatible with number of sub grids.")
 
-        # allocate arrays for constructing dataset
-        self.darcy0 = wp.zeros(self.dim, dtype=float, device=self.device)
-        self.darcy1 = wp.zeros(self.dim, dtype=float, device=self.device)
-        self.permeability = wp.zeros(self.dim, dtype=float, device=self.device)
-        self.rand_fourier = wp.zeros(self.fourier_dim, dtype=float, device=self.device)
-        self.inf_residual = wp.zeros([1], dtype=float, device=self.device)
         self.train_path = train_path
-        self.downsample = 5
-        self.r = self.downsample
-        self.h = int(((421 - 1) / self.r) + 1)
-        self.s = self.h
-        # print(f"=============={self.s}===============")
+        self.native_resolution = 421  # Native grid size
+
+        # Calculate downsampling factor
+        if (self.native_resolution - 1) % (self.resolution - 1) != 0:
+            raise ValueError(
+                f"Resolution {self.resolution} is not achievable by strided sampling from native resolution {self.native_resolution}."
+            )
+        self.r = (self.native_resolution - 1) // (self.resolution - 1)
+        self.s = self.resolution
         self.dx = 1.0 / self.s
-
         # Output tenors
         self.output_k = None
         self.output_p = None
@@ -217,10 +197,9 @@ def __init__(
         self.is_test = is_test
 
         if not self.is_test:
-            n_train = 1000
+            self.n_train = 1024
         else:
-            n_train = 200
-        self.n_train = n_train
+            self.n_train = 200
 
         if self.train_path is not None:
             self.__get_data__()
@@ -233,28 +212,58 @@ def __init__(
             self.y_train = self.y_normalizer.encode(self.y_train)
         else:
             self.x_train = x_normalizer.encode(self.x_train)
+            self.y_train = y_normalizer.encode(self.y_train)
 
+    @profile
     def __get_normalizer__(self):
         return self.x_normalizer, self.y_normalizer
 
+    @profile
     def __get_data__(self):
+
+        if self.train_path.endswith(".mat"):
+            data_dict = scio.loadmat(self.train_path)
+        elif self.train_path.endswith(".npz"):
+            data_dict = np.load(self.train_path)
+
+        # Extract data from dicts:
+        self.x_train = data_dict["coeff"]
+        self.y_train = data_dict["sol"]
+
         x = np.linspace(0, 1, self.s)
         y = np.linspace(0, 1, self.s)
         x, y = np.meshgrid(x, y)
         pos = np.c_[x.ravel(), y.ravel()]
-        pos = torch.tensor(pos, dtype=torch.float).unsqueeze(0).cuda()
-        self.x_train = scio.loadmat(self.train_path)["coeff"][
-            : self.n_train, :: self.r, :: self.r
-        ][:, : self.s, : self.s]
+        pos = torch.tensor(pos, dtype=torch.float).cuda()
+
+        # Downsampling logic
+        if self.r > 1:
+            # Downsample by slicing
+            self.x_train = self.x_train[: self.n_train, :: self.r, :: self.r][
+                :, : self.s, : self.s
+            ]
+            self.y_train = self.y_train[: self.n_train, :: self.r, :: self.r][
+                :, : self.s, : self.s
+            ]
+        else:
+            # No downsampling, use full resolution
+            self.x_train = self.x_train[: self.n_train, : self.s, : self.s]
+            self.y_train = self.y_train[: self.n_train, : self.s, : self.s]
+
+        # Flatten them:
         self.x_train = self.x_train.reshape(self.n_train, -1)
-        self.x_train = torch.from_numpy(self.x_train).float().cuda()
-        self.y_train = scio.loadmat(self.train_path)["sol"][
-            : self.n_train, :: self.r, :: self.r
-        ][:, : self.s, : self.s]
         self.y_train = self.y_train.reshape(self.n_train, -1)
-        self.y_train = torch.from_numpy(self.y_train).float().cuda()
-        self.pos_train = pos.repeat(self.n_train, 1, 1)
 
+        self.x_train = torch.from_numpy(self.x_train).float().cuda()
+        self.y_train = torch.from_numpy(self.y_train).float().cuda()
+        # Why are we repeating the postion?
+        # print(f"pos shape: {pos.shape}")
+        self.pos_train = pos
+        self.pos_train_batched = pos.repeat(self.batch_size, 1, 1).cuda()
+        # print(f"pos shape post repeat: {self.pos_train.shape}")
+        # self.pos_train = pos
+
+    @profile
     def __iter__(self):
         """
         Yields
@@ -263,13 +272,17 @@ def __iter__(self):
             Infinite iterator that returns a batch of (permeability, darcy pressure)
             fields of size [batch, resolution, resolution]
         """
-        # infinite generator
+
         while True:
-            idx = np.random.choice(200, self.batch_size)
+            # Sample batch_size indices from this rank's shard
+            idx = np.random.choice(self.n_train, self.batch_size)
+            # All tensors are already on GPU, so no .cuda() needed
             x = self.x_train[idx]
             y = self.y_train[idx]
-            pos = self.pos_train[idx]
-            yield pos, x, y
+            yield self.pos_train_batched, x, y
+
+    def __getitem__(self, idx):
+        return self.pos_train, self.x_train[idx], self.y_train[idx]
 
     def __len__(self):
-        return self.n_train // self.batch_size
+        return self.n_train
diff --git a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
index 76d2599d10..e218b67fed 100644
--- a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
+++ b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
@@ -14,68 +14,43 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-
+# Configuration imports:
 import hydra
-from omegaconf import DictConfig
-from math import ceil
+from omegaconf import DictConfig, OmegaConf
+import json
 
-from torch.nn import MSELoss
-from utils.testloss import TestLoss
-from torch.optim import Adam, lr_scheduler, AdamW
+from math import ceil
 
-from physicsnemo.models.transolver import Transolver
-from physicsnemo.distributed import DistributedManager
-from physicsnemo.utils import StaticCaptureTraining, StaticCaptureEvaluateNoGrad
-from physicsnemo.launch.utils import load_checkpoint, save_checkpoint
-from physicsnemo.launch.logging import PythonLogger, LaunchLogger
-from physicsnemo.launch.logging.mlflow import initialize_mlflow
+# Base PyTorch imports:
+import torch
+import torch.distributed as dist
 
-from darcy_datapipe_fix import Darcy2D_fix
-from validator_fix import GridValidator
 
+from torch.optim import lr_scheduler, AdamW
+from torch.nn.parallel import DistributedDataParallel as DDP
 
-class UnitTransformer:
-    """Unit transformer class for normalizing and denormalizing data."""
+# PyTorch Data tools
+from torch.utils.data import DataLoader, DistributedSampler
 
-    def __init__(self, X):
-        self.mean = X.mean(dim=(0, 1), keepdim=True)
-        self.std = X.std(dim=(0, 1), keepdim=True) + 1e-8
+from torch.utils.tensorboard import SummaryWriter
 
-    def to(self, device):
-        self.mean = self.mean.to(device)
-        self.std = self.std.to(device)
-        return self
+from utils.testloss import TestLoss
 
-    def cuda(self):
-        self.mean = self.mean.cuda()
-        self.std = self.std.cuda()
+# Model imports from PhysicsNeMo
+from physicsnemo.models.transolver import Transolver
+from physicsnemo.distributed import DistributedManager
 
-    def cpu(self):
-        self.mean = self.mean.cpu()
-        self.std = self.std.cpu()
+from physicsnemo.launch.utils import load_checkpoint, save_checkpoint
+from physicsnemo.launch.logging import PythonLogger, RankZeroLoggingWrapper
 
-    def encode(self, x):
-        x = (x - self.mean) / (self.std)
-        return x
+from darcy_datapipe_fix import Darcy2D_fix
+from validator_fix import GridValidator
 
-    def decode(self, x):
-        return x * self.std + self.mean
+from physicsnemo.utils.profiling import Profiler
+from contextlib import nullcontext
+import time  # <-- Add this import for timing
 
-    def transform(self, X, inverse=True, component="all"):
-        if component == "all" or "all-reduce":
-            if inverse:
-                orig_shape = X.shape
-                return (X * (self.std - 1e-8) + self.mean).view(orig_shape)
-            else:
-                return (X - self.mean) / self.std
-        else:
-            if inverse:
-                orig_shape = X.shape
-                return (
-                    X * (self.std[:, component] - 1e-8) + self.mean[:, component]
-                ).view(orig_shape)
-            else:
-                return (X - self.mean[:, component]) / self.std[:, component]
+prof = Profiler()
 
 
 def count_parameters(model):
@@ -85,30 +60,127 @@ def count_parameters(model):
             continue
         params = parameter.numel()
         total_params += params
-    print(f"Total Trainable Params: {total_params}")
     return total_params
 
 
+def forward_train_full_loop(
+    model, loss_fun, optimizer, pos, x, y, y_normalizer, context, scaler=None
+):
+    """
+    Forward and backward pass for one iteration, with optional mixed precision training.
+    Returns the loss for this minibatch
+    """
+    with context:
+        pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
+        pred = y_normalizer.decode(pred)
+        loss = loss_fun(pred, y)
+    if scaler is not None:
+        scaler.scale(loss).backward()
+        scaler.step(optimizer)
+        scaler.update()
+    else:
+        loss.backward()
+        optimizer.step()
+        optimizer.zero_grad()
+    return loss
+
+
+def train_epoch(
+    model,
+    optimizer,
+    scheduler,
+    train_dataloader,
+    loss_fun,
+    y_normalizer,
+    context,
+    scaler,
+):
+    """
+    One epoch of training.  Returns the loss from the last minibatch used, averaged across replicas.
+
+    """
+    for i, batch in enumerate(train_dataloader):
+        pos, x, y = batch
+        loss = forward_train_full_loop(
+            model, loss_fun, optimizer, pos, x, y, y_normalizer, context, scaler
+        )
+        scheduler.step()
+
+    # At the end of the epoch, reduce the last local loss if needed:
+    dm = DistributedManager()
+    if dm.world_size > 1:
+        dist.all_reduce(loss.detach(), op=dist.ReduceOp.SUM)
+        loss = loss / dm.world_size
+
+    return loss
+
+
+def val_epoch(model, test_dataloader, loss_fun, y_normalizer):
+    """
+    One epoch of validation.  Returns the loss averaged across the entire validation set.
+    """
+    val_loss = None
+    RL2 = None
+    for i, batch in enumerate(test_dataloader):
+        pos, x, y = batch
+        with torch.no_grad():
+            pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
+            pred = y_normalizer.decode(pred)
+            loss = loss_fun(pred, y)
+
+            # --- Relative L2 calculation ---
+            # print(f"pred shape: {pred.shape}")
+            # print(f"y shape: {y.shape}")
+            rel_l2 = torch.norm(pred.reshape(y.shape) - y) / torch.norm(y)
+            if RL2 is None:
+                RL2 = rel_l2
+            else:
+                RL2 += rel_l2
+            if val_loss is None:
+                val_loss = loss
+            else:
+                val_loss += loss
+    val_loss = val_loss / len(test_dataloader)
+
+    dm = DistributedManager()
+    if dm.world_size > 1:
+        dist.all_reduce(val_loss, op=dist.ReduceOp.SUM)
+        dist.all_reduce(RL2, op=dist.ReduceOp.SUM)
+        val_loss = val_loss / dm.world_size
+        RL2 = RL2 / dm.world_size
+    return val_loss, pred, y, RL2
+
+
 @hydra.main(version_base="1.3", config_path=".", config_name="config_fix.yaml")
 def darcy_trainer(cfg: DictConfig) -> None:
     """Training for the 2D Darcy flow benchmark problem."""
+    ########################################################################
+    # Initialize distributed tools
+    ########################################################################
     DistributedManager.initialize()  # Only call this once in the entire script!
-    dist = DistributedManager()  # call if required elsewhere
-
-    # initialize monitoring
-    log = PythonLogger(name="darcy_transolver")
-    log.file_logging()
-    initialize_mlflow(
-        experiment_name=f"Darcy_Transolver",
-        experiment_desc=f"training a Transformer-based PDE solver for the Darcy problem",
-        run_name=f"Darcy Transolver training",
-        run_desc=f"training Transolver for Darcy",
-        user_name="Haixu Wu, Huakun Luo, Haowen Wang",
-        mode="offline",
-    )
-    LaunchLogger.initialize(use_mlflow=True)  # PhysicsNeMo launch logger
-
-    # define model, loss, optimiser, scheduler, data loader
+    dm = DistributedManager()  # call if required elsewhere
+
+    ########################################################################
+    # Initialize monitoring and logging
+    ########################################################################
+    logger = RankZeroLoggingWrapper(PythonLogger(name="darcy_transolver"), dm)
+    logger.file_logging()
+
+    # === TensorBoard SummaryWriter ===
+    # Only rank 0 writes logs to avoid duplication in DDP
+    writer = None
+    if dm.rank == 0:
+        log_dir = f"{cfg.output_dir}/runs/{cfg.run_id}"
+        writer = SummaryWriter(log_dir=log_dir)
+
+    ########################################################################
+    # Print the configuration to log
+    ########################################################################
+    logger.info(json.dumps(OmegaConf.to_container(cfg), indent=4))
+
+    ########################################################################
+    # define model
+    ########################################################################
     model = Transolver(
         space_dim=cfg.model.space_dim,
         n_layers=cfg.model.n_layers,
@@ -123,42 +195,73 @@ def darcy_trainer(cfg: DictConfig) -> None:
         slice_num=cfg.model.slice_num,
         ref=cfg.model.ref,
         unified_pos=cfg.model.unified_pos,
-        H=cfg.training.resolution,
-        W=cfg.training.resolution,
-    ).to(dist.device)
-    count_parameters(model)
-    loss_fun = TestLoss(size_average=False)
+        H=cfg.data.resolution,
+        W=cfg.data.resolution,
+        use_te=cfg.model.transformer_engine,
+    ).to(dm.device)
+
+    if dm.world_size > 1:
+        model = DDP(model, device_ids=[dm.rank])
+
+    n_params = count_parameters(model)
+    logger.info(f"Total Trainable Params: {n_params}")
+
+    ########################################################################
+    # define loss and optimizer
+    ########################################################################
+    loss_fun = TestLoss(size_average=True)
     optimizer = AdamW(
         model.parameters(),
         lr=cfg.scheduler.initial_lr,
         weight_decay=cfg.scheduler.weight_decay,
     )
-    # scheduler = lr_scheduler.LambdaLR(
-    #     optimizer, lr_lambda=lambda step: cfg.scheduler.decay_rate**step
-    # )
-
-    norm_vars = cfg.normaliser
-    normaliser = {
-        "permeability": (norm_vars.permeability.mean, norm_vars.permeability.std_dev),
-        "darcy": (norm_vars.darcy.mean, norm_vars.darcy.std_dev),
-    }
-    # train_dataloader = Darcy2D_fix(
-    #     resolution=cfg.training.resolution,
-    #     batch_size=cfg.training.batch_size,
-    #     normaliser=normaliser,
-    #     train_path="/data/fno/piececonst_r421_N1024_smooth1.mat",
-    #     is_test=False,
-    # )
-    train_dataloader = Darcy2D_fix(
-        resolution=cfg.training.resolution,
-        batch_size=cfg.training.batch_size,
-        normaliser=normaliser,
-        train_path="/data/fno/piececonst_r421_N1024_smooth1.mat",
+
+    ########################################################################
+    # Create the data pipes and samplers
+    ########################################################################
+
+    train_datapipe = Darcy2D_fix(
+        resolution=cfg.data.resolution,
+        batch_size=cfg.data.batch_size,
+        train_path=cfg.data.train_path,
         is_test=False,
     )
+    # Sampler ensures disjoint instances on each rank
+    train_sampler = DistributedSampler(
+        train_datapipe, num_replicas=dm.world_size, rank=dm.rank, shuffle=True
+    )
+    # DataLoader handles the batching
+    train_dataloader = DataLoader(
+        train_datapipe,
+        batch_size=cfg.data.batch_size // dm.world_size,
+        sampler=train_sampler,
+        drop_last=True,
+    )
+    # Reuse the train normalizer for the test data:
+    # (The normalizer puts the inputs and targets to mean 0, std=1.0)
+    x_normalizer, y_normalizer = train_datapipe.__get_normalizer__()
+
+    test_datapipe = Darcy2D_fix(
+        resolution=cfg.data.resolution,
+        batch_size=cfg.data.batch_size,
+        train_path=cfg.data.test_path,
+        is_test=True,
+        x_normalizer=x_normalizer,
+        y_normalizer=y_normalizer,
+    )
+    test_sampler = DistributedSampler(
+        test_datapipe, num_replicas=dm.world_size, rank=dm.rank, shuffle=False
+    )
+    test_dataloader = DataLoader(
+        test_datapipe,
+        batch_size=cfg.data.batch_size // dm.world_size,
+        sampler=test_sampler,
+        drop_last=True,
+    )
+
     # calculate steps per pseudo epoch
     steps_per_pseudo_epoch = ceil(
-        cfg.training.pseudo_epoch_sample_size / cfg.training.batch_size
+        cfg.training.pseudo_epoch_sample_size / cfg.data.batch_size
     )
 
     scheduler = lr_scheduler.OneCycleLR(
@@ -168,106 +271,145 @@ def darcy_trainer(cfg: DictConfig) -> None:
         epochs=cfg.training.max_pseudo_epochs,
     )
 
-    x_normalizer, y_normalizer = train_dataloader.__get_normalizer__()
-
-    test_dataloader = Darcy2D_fix(
-        resolution=cfg.training.resolution,
-        batch_size=cfg.training.batch_size,
-        normaliser=normaliser,
-        train_path="/data/fno/piececonst_r421_N1024_smooth2.mat",
-        is_test=True,
-        x_normalizer=x_normalizer,
-    )
-
-    validator = GridValidator(loss_fun=TestLoss(size_average=False), norm=y_normalizer)
+    validator = GridValidator(output_dir=f"{cfg.output_dir}/runs/{cfg.run_id}/plots")
 
     ckpt_args = {
-        "path": f"./checkpoints",
+        "path": f"{cfg.output_dir}/runs/{cfg.run_id}/checkpoints",
         "optimizer": optimizer,
         "scheduler": scheduler,
         "models": model,
     }
-    loaded_pseudo_epoch = load_checkpoint(device=dist.device, **ckpt_args)
+    loaded_pseudo_epoch = load_checkpoint(device=dm.device, **ckpt_args)
 
-    validation_iters = ceil(cfg.validation.sample_size / cfg.training.batch_size)
-    log_args = {
-        "name_space": "train",
-        "num_mini_batch": steps_per_pseudo_epoch,
-        "epoch_alert_freq": 1,
-    }
-    if cfg.training.pseudo_epoch_sample_size % cfg.training.batch_size != 0:
-        log.warning(
+    validation_iters = ceil(cfg.validation.sample_size / cfg.data.batch_size)
+
+    if cfg.training.pseudo_epoch_sample_size % cfg.data.batch_size != 0:
+        logger.warning(
             f"increased pseudo_epoch_sample_size to multiple of \
-                      batch size: {steps_per_pseudo_epoch*cfg.training.batch_size}"
+                      batch size: {steps_per_pseudo_epoch*cfg.data.batch_size}"
         )
-    if cfg.validation.sample_size % cfg.training.batch_size != 0:
-        log.warning(
+    if cfg.validation.sample_size % cfg.data.batch_size != 0:
+        logger.warning(
             f"increased validation sample size to multiple of \
-                      batch size: {validation_iters*cfg.training.batch_size}"
+                      batch size: {validation_iters*cfg.data.batch_size}"
         )
 
-    # define forward passes for training and inference
-    @StaticCaptureTraining(
-        model=model, optim=optimizer, logger=log, use_amp=False, use_graphs=False
-    )
-    def forward_train(pos, x, y, y_normalizer):
-        pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
-        pred = y_normalizer.decode(pred)
-        loss = loss_fun(pred, y)
-        return loss
-
-    @StaticCaptureEvaluateNoGrad(
-        model=model, logger=log, use_amp=False, use_graphs=False
-    )
-    def forward_eval(pos, x, y, y_normalizer):
-        pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
-        return y_normalizer.decode(pred)
+    # Initialize GradScaler for mixed precision training
+    if cfg.precision == "fp16":
+        context = torch.amp.autocast(device_type="cuda", dtype=torch.float16)
+        scaler = torch.amp.GradScaler("cuda")
+    elif cfg.precision == "bf16":
+        context = torch.amp.autocast(device_type="cuda", dtype=torch.bfloat16)
+        scaler = None
+    else:
+        context = nullcontext()
+        scaler = None
 
     if loaded_pseudo_epoch == 0:
-        log.success("Training started...")
+        logger.success("Training started...")
     else:
-        log.warning(f"Resuming training from pseudo epoch {loaded_pseudo_epoch+1}.")
-
-    for pseudo_epoch in range(
-        max(1, loaded_pseudo_epoch + 1), cfg.training.max_pseudo_epochs + 1
-    ):
-        # Wrap epoch in launch logger for console / MLFlow logs
-        with LaunchLogger(**log_args, epoch=pseudo_epoch) as logger:
-            for _, batch in zip(range(steps_per_pseudo_epoch), train_dataloader):
-                loss = forward_train(*batch, y_normalizer)
-                logger.log_minibatch({"loss": loss.detach() / cfg.training.batch_size})
-                scheduler.step()
-            logger.log_epoch({"Learning Rate": optimizer.param_groups[0]["lr"]})
-
-        # save checkpoint
-        if pseudo_epoch % cfg.training.rec_results_freq == 0:
-            save_checkpoint(**ckpt_args, epoch=pseudo_epoch)
-
-        # validation step
-        if pseudo_epoch % cfg.validation.validation_pseudo_epochs == 0:
-            with LaunchLogger("valid", epoch=pseudo_epoch) as logger:
-                total_loss = 0.0
-                for _, batch in zip(range(validation_iters), test_dataloader):
-                    val_loss = validator.compare(
-                        batch[2],
-                        forward_eval(*batch, y_normalizer),
+        logger.warning(f"Resuming training from pseudo epoch {loaded_pseudo_epoch+1}.")
+
+    # Get the first batch of the test dataset for plotting
+
+    with prof:
+
+        for pseudo_epoch in range(
+            max(1, loaded_pseudo_epoch + 1), cfg.training.max_pseudo_epochs + 1
+        ):
+            # --- TRAINING ---
+            train_start = time.time()
+            loss = train_epoch(
+                model,
+                optimizer,
+                scheduler,
+                train_dataloader,
+                loss_fun,
+                y_normalizer,
+                context,
+                scaler,
+            )
+            train_time = time.time() - train_start
+
+            # After training epoch, e.g. after loss, train_time, optimizer, etc. are available:
+            if torch.cuda.is_available():
+                gpu_mem_reserved = torch.cuda.memory_reserved() / 1024**3
+            else:
+                gpu_mem_reserved = 0
+
+            lr = optimizer.param_groups[0]["lr"]
+
+            header = "mode\tEpoch\tloss\ttime\tLR\t\tGPU_mem"
+            values = f"train\t{pseudo_epoch}\t{loss.item():.4f}\t{train_time:.2f}\t{lr:.4e}\t{gpu_mem_reserved:.2f}"
+
+            log_string = f"\n{header}\n{values}"
+            logger.info(log_string)
+
+            # --- TensorBoard logging (only on rank 0) ---
+            if dm.rank == 0 and writer is not None:
+                # Images/sec/GPU: (num images processed in train_epoch) / train_time / num_gpus
+                # Each batch processes batch_size // world_size images, for steps_per_pseudo_epoch steps
+                images_per_epoch = len(train_dataloader) * (
+                    cfg.data.batch_size // dm.world_size
+                )
+                images_per_sec_per_gpu = images_per_epoch / train_time
+
+                writer.add_scalar("loss/train", loss.item(), pseudo_epoch)
+                writer.add_scalar("time_per_epoch/train", train_time, pseudo_epoch)
+                writer.add_scalar(
+                    "images_per_sec_per_gpu/train", images_per_sec_per_gpu, pseudo_epoch
+                )
+                writer.add_scalar("learning_rate/train", lr, pseudo_epoch)
+
+            # save checkpoint
+            if pseudo_epoch % cfg.training.rec_results_freq == 0 and dm.rank == 0:
+                save_checkpoint(**ckpt_args, epoch=pseudo_epoch)
+
+            # --- VALIDATION ---
+            if pseudo_epoch % cfg.validation.validation_pseudo_epochs == 0:
+                val_start = time.time()
+                val_loss, pred, y, RL2 = val_epoch(
+                    model, test_dataloader, loss_fun, y_normalizer
+                )
+                val_time = time.time() - val_start
+
+                header = "mode\tEpoch\tloss\tRL2\ttime"
+                values = f"val\t{pseudo_epoch}\t{val_loss.item():.4f}\t{RL2.item():.4f}\t{val_time:.2f}"
+
+                log_string = f"\n{header}\n{values}"
+                logger.info(log_string)
+
+                # --- TensorBoard logging (only on rank 0) ---
+                if dm.rank == 0 and writer is not None:
+                    # Validation images/sec/GPU
+                    val_images = validation_iters * (
+                        cfg.data.batch_size // dm.world_size
+                    )
+                    val_images_per_sec_per_gpu = val_images / val_time
+                    writer.add_scalar("loss/val", val_loss.item(), pseudo_epoch)
+                    writer.add_scalar("RL2/val", RL2.item(), pseudo_epoch)
+                    writer.add_scalar("time_per_epoch/val", val_time, pseudo_epoch)
+                    writer.add_scalar(
+                        "images_per_sec_per_gpu/val",
+                        val_images_per_sec_per_gpu,
                         pseudo_epoch,
-                        logger,
                     )
-                    total_loss += val_loss
-                logger.log_epoch(
-                    {
-                        "Validation error": total_loss
-                        / (validation_iters * cfg.training.batch_size)
-                    }
-                )
+
+                if dm.rank == 0:
+                    validator.make_plot(pred, y, pseudo_epoch, test_datapipe.s)
 
         # update learning rate
         # if pseudo_epoch % cfg.scheduler.decay_pseudo_epochs == 0:
 
-    save_checkpoint(**ckpt_args, epoch=cfg.training.max_pseudo_epochs)
-    log.success("Training completed *yay*")
+    if dm.rank == 0 and writer is not None:
+        writer.close()
+    logger.success("Training completed *yay*")
 
 
 if __name__ == "__main__":
+    # prof.enable("line_profile")
+    # prof.enable("torch")
+    # prof.initialize()
     darcy_trainer()
+
+    # prof.finalize()
diff --git a/examples/cfd/darcy_transolver/validator_fix.py b/examples/cfd/darcy_transolver/validator_fix.py
index 658f9cfff4..2714179b8a 100644
--- a/examples/cfd/darcy_transolver/validator_fix.py
+++ b/examples/cfd/darcy_transolver/validator_fix.py
@@ -16,7 +16,8 @@
 
 import matplotlib.pyplot as plt
 from torch import FloatTensor
-from physicsnemo.launch.logging import LaunchLogger
+import threading
+import os
 
 
 class GridValidator:
@@ -37,21 +38,50 @@ class GridValidator:
 
     def __init__(
         self,
-        loss_fun,
-        norm,
         font_size: float = 28.0,
+        output_dir: str = "./plots/",
     ):
-        self.norm = norm
-        self.criterion = loss_fun
         self.font_size = font_size
         self.headers = ("true", "prediction", "error")
+        self._plot_thread = None
+        self.output_dir = output_dir
+        os.makedirs(self.output_dir, exist_ok=True)
 
-    def compare(
+    def plot_figure(
+        self, target: FloatTensor, prediction: FloatTensor, step: int, resolution: int
+    ):
+
+        target = target.cpu().numpy().reshape(-1, resolution, resolution)[0, :, :]
+        prediction = (
+            prediction.reshape(-1, resolution, resolution)
+            .detach()
+            .cpu()
+            .numpy()[0, :, :]
+        )
+
+        plt.close("all")
+        plt.rcParams.update({"font.size": self.font_size})
+        fig, ax = plt.subplots(1, 3, figsize=(15 * 3, 15), sharey=True)
+        im = []
+        im.append(ax[0].imshow(target))
+        im.append(ax[1].imshow(prediction))
+        im.append(ax[2].imshow((prediction - target)))
+
+        for ii in range(len(im)):
+            fig.colorbar(im[ii], ax=ax[ii], location="bottom", fraction=0.046, pad=0.04)
+            ax[ii].set_title(self.headers[ii])
+
+        plt.savefig(f"{self.output_dir}/validation_step_{step:03d}.png")
+
+    def _plot_figure_thread(self, target, prediction, step, resolution):
+        self.plot_figure(target, prediction, step, resolution)
+
+    def make_plot(
         self,
         prediction: FloatTensor,
         target: FloatTensor,
         step: int,
-        logger: LaunchLogger,
+        resolution: int,
     ) -> float:
         """compares model output, target and plots everything
 
@@ -73,25 +103,16 @@ def compare(
         float
             validation error
         """
-        loss = self.criterion(prediction, target)
-        # print(f"target.shape: {target.shape}, prediction.shape: {prediction.shape}")
-        print("logger begin")
-        target = target.cpu().numpy()[0, :, :]
-        prediction = prediction.reshape(-1, 85, 85).detach().cpu().numpy()[0, :, :]
-
-        plt.close("all")
-        plt.rcParams.update({"font.size": self.font_size})
-        fig, ax = plt.subplots(1, 3, figsize=(15 * 3, 15), sharey=True)
-        im = []
-        im.append(ax[0].imshow(target))
-        im.append(ax[1].imshow(prediction))
-        im.append(ax[2].imshow((prediction - target)))
 
-        for ii in range(len(im)):
-            fig.colorbar(im[ii], ax=ax[ii], location="bottom", fraction=0.046, pad=0.04)
-            ax[ii].set_title(self.headers[ii])
+        # Wait for previous plot thread if still running
+        if self._plot_thread is not None and self._plot_thread.is_alive():
+            self._plot_thread.join()
 
-        logger.log_figure(figure=fig, artifact_file=f"validation_step_{step:03d}.png")
-        print("logger finished")
+        # Start new plot thread
+        self._plot_thread = threading.Thread(
+            target=self._plot_figure_thread,
+            args=(target, prediction, step, resolution),
+        )
+        self._plot_thread.start()
 
-        return loss
+        return
diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index c4b7807d7c..0a169a97f4 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -52,7 +52,7 @@ class Physics_Attention_Base(nn.Module):
 
     """
 
-    def __init__(self, dim, heads, dim_head, dropout, slice_num):
+    def __init__(self, dim, heads, dim_head, dropout, slice_num, use_te):
         super().__init__()
         inner_dim = dim_head * heads
         self.dim_head = dim_head
@@ -68,17 +68,22 @@ def __init__(self, dim, heads, dim_head, dropout, slice_num):
         for l_i in [self.in_project_slice]:
             torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
 
-        self.to_q = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_k = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_v = nn.Linear(dim_head, dim_head, bias=False)
-
-        # # These are used in the transformer engine pass function:
-        # self.qkv_project = nn.Linear(dim_head, 3 * dim_head, bias=False)
-        # self.attn_fn = te.DotProductAttention(num_attention_heads=self.heads,
-        #                                       kv_channels= self.dim_head,
-        #                                       attention_dropout=dropout,
-        #                                       qkv_format="bshd",
-        #                                       softmax_scale=self.scale)
+        if not use_te:
+            self.to_q = nn.Linear(dim_head, dim_head, bias=False)
+            self.to_k = nn.Linear(dim_head, dim_head, bias=False)
+            self.to_v = nn.Linear(dim_head, dim_head, bias=False)
+        else:
+            # These are used in the transformer engine pass function:
+            self.qkv_project = nn.Linear(dim_head, 3 * dim_head, bias=False)
+            self.attn_fn = te.DotProductAttention(
+                num_attention_heads=self.heads,
+                kv_channels=self.dim_head,
+                attention_dropout=dropout,
+                qkv_format="bshd",
+                softmax_scale=self.scale,
+            )
+
+        self.use_te = use_te
 
         self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
 
@@ -114,12 +119,16 @@ def compute_slices_from_projections(
             - It then aggregates the latent features (fx) for each slice using these weights.
             - The aggregated features are normalized by the sum of weights for numerical stability.
         """
+
         # Project the latent space vectors on to the weight computation space,
         # and compute a temperature adjusted softmax.
+        clamped_temp = torch.clamp(self.temperature, min=0.1, max=5).to(
+            slice_projections.dtype
+        )
         slice_weights = nn.functional.softmax(
-            slice_projections / torch.clamp(self.temperature, min=0.1, max=5), dim=-1
+            slice_projections / clamped_temp, dim=-1
         )  # [Batch, N_heads, N_tokens, Slice_num]
-
+        slice_weights = slice_weights.to(slice_projections.dtype)
         # Average the slices over the token dimension
         slice_norm = slice_weights.sum(2)  # [Batch, N_heads, Slice_num]
 
@@ -146,7 +155,7 @@ def compute_slice_attention(self, slice_tokens: torch.Tensor) -> torch.Tensor:
 
         return out_slice_token
 
-    def compute_slice_attention2(self, slice_tokens: torch.Tensor) -> torch.Tensor:
+    def compute_slice_attention_te(self, slice_tokens: torch.Tensor) -> torch.Tensor:
         """
         TE implementation of slice attention
         """
@@ -162,7 +171,7 @@ def compute_slice_attention2(self, slice_tokens: torch.Tensor) -> torch.Tensor:
 
         return out_slice_token2
 
-    def compute_slice_attention3(self, slice_tokens: torch.Tensor) -> torch.Tensor:
+    def compute_slice_attention_sdpa(self, slice_tokens: torch.Tensor) -> torch.Tensor:
         """
         Torch SDPA implementation of slice attention
         """
@@ -233,9 +242,11 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
         # slice_tokens has shape  [Batch, N_heads, N_tokens, head_dim]
 
         # Apply attention to the slice tokens
-        # out_slice_token = self.compute_slice_attention(slice_tokens)
-        # out_slice_token = self.compute_slice_attention2(slice_tokens)
-        out_slice_token = self.compute_slice_attention3(slice_tokens)
+        if self.use_te:
+            out_slice_token = self.compute_slice_attention_te(slice_tokens)
+        else:
+            out_slice_token = self.compute_slice_attention_sdpa(slice_tokens)
+
         # Shape unchanged
 
         # Deslice:
@@ -251,8 +262,10 @@ class Physics_Attention_Irregular_Mesh_2(Physics_Attention_Base):
     Specialization of PhysicsAttention to Irregular Meshes
     """
 
-    def __init__(self, dim, heads=8, dim_head=64, dropout=0.0, slice_num=64):
-        super().__init__(dim, heads, dim_head, dropout, slice_num)
+    def __init__(
+        self, dim, heads=8, dim_head=64, dropout=0.0, slice_num=64, use_te=True
+    ):
+        super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
         inner_dim = dim_head * heads
 
         self.in_project_x = nn.Linear(dim, inner_dim)
@@ -288,8 +301,9 @@ def __init__(
         H=101,
         W=31,
         kernel=3,
+        use_te=True,
     ):  # kernel=3):
-        super().__init__(dim, heads, dim_head, dropout, slice_num)
+        super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
 
         inner_dim = dim_head * heads
         self.H = H
@@ -341,8 +355,9 @@ def __init__(
         W=32,
         D=32,
         kernel=3,
+        use_te=True,
     ):
-        super().__init__(dim, heads, dim_head, dropout, slice_num)
+        super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
 
         inner_dim = dim_head * heads
         self.H = H
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index 875c31b350..be3f93341f 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -34,7 +34,13 @@
 import numpy as np
 import torch
 import torch.nn as nn
-import transformer_engine.pytorch as te
+
+try:
+    import transformer_engine.pytorch as te
+
+    TE_AVAILABLE = True
+except ImportError:
+    TE_AVAILABLE = False
 
 import physicsnemo  # noqa: F401 for docs
 
@@ -82,7 +88,6 @@ def __init__(self, n_input, n_hidden, n_output, n_layers=1, act="gelu", res=True
         )
 
     def forward(self, x):
-        # print(x.shape)
         x = self.linear_pre(x)
         for i in range(self.n_layers):
             if self.res:
@@ -108,10 +113,20 @@ def __init__(
         slice_num=32,
         H=85,
         W=85,
+        use_te=True,
     ):
         super().__init__()
+
+        if use_te and not TE_AVAILABLE:
+            raise ImportError(
+                "Transformer Engine is not installed. Please install it with `pip install transformer-engine`."
+            )
+
         self.last_layer = last_layer
-        self.ln_1 = te.LayerNorm(hidden_dim)
+        if use_te:
+            self.ln_1 = te.LayerNorm(hidden_dim)
+        else:
+            self.ln_1 = nn.LayerNorm(hidden_dim)
         self.Attn = Physics_Attention_Structured_Mesh_2D(
             hidden_dim,
             heads=num_heads,
@@ -120,28 +135,36 @@ def __init__(
             slice_num=slice_num,
             H=H,
             W=W,
+            use_te=use_te,
         )
 
-        self.ln_mlp1 = te.LayerNormMLP(
-            hidden_size=hidden_dim,
-            ffn_hidden_size=hidden_dim * mlp_ratio,
-        )
-
-        # self.ln_2 = te.LayerNorm(hidden_dim)
-        # self.mlp = MLP(
-        #     hidden_dim,
-        #     hidden_dim * mlp_ratio,
-        #     hidden_dim,
-        #     n_layers=0,
-        #     res=False,
-        #     act=act,
-        # )
-        if self.last_layer:
-            self.ln_mlp2 = te.LayerNormLinear(
-                in_features=hidden_dim, out_features=out_dim
+        if use_te:
+            self.ln_mlp1 = te.LayerNormMLP(
+                hidden_size=hidden_dim,
+                ffn_hidden_size=hidden_dim * mlp_ratio,
             )
-            # self.ln_3 = te.LayerNorm(hidden_dim)
-            # self.mlp2 = nn.Linear(hidden_dim, out_dim)
+        else:
+            self.ln_mlp1 = nn.Sequential(
+                nn.LayerNorm(hidden_dim),
+                MLP(
+                    hidden_dim,
+                    hidden_dim * mlp_ratio,
+                    hidden_dim,
+                    n_layers=0,
+                    res=False,
+                    act=act,
+                ),
+            )
+        if self.last_layer:
+            if use_te:
+                self.ln_mlp2 = te.LayerNormLinear(
+                    in_features=hidden_dim, out_features=out_dim
+                )
+            else:
+                self.ln_mlp2 = nn.Sequential(
+                    nn.LayerNorm(hidden_dim),
+                    nn.Linear(hidden_dim, out_dim),
+                )
 
     def forward(self, fx):
         fx = self.Attn(self.ln_1(fx)) + fx
@@ -172,6 +195,7 @@ def __init__(
         unified_pos=False,
         H=85,
         W=85,
+        use_te=True,
     ):
         super().__init__()
         self.__name__ = "Transolver_2D"
@@ -219,14 +243,15 @@ def __init__(
                     H=H,
                     W=W,
                     last_layer=(_ == n_layers - 1),
+                    use_te=use_te,
                 )
                 for _ in range(n_layers)
             ]
         )
         self.initialize_weights()
-        self.placeholder = nn.Parameter(
-            (1 / (n_hidden)) * torch.rand(n_hidden, dtype=torch.float)
-        )
+        # self.placeholder = nn.Parameter(
+        #     (1 / (n_hidden)) * torch.rand(n_hidden, dtype=torch.float)
+        # )
 
     def initialize_weights(self):
         self.apply(self._init_weights)
@@ -270,7 +295,6 @@ def get_grid(self, batchsize=1):
     def forward(self, x, fx, T=None):
         with torch.autograd.profiler.record_function("prepeocess"):
             if self.unified_pos:
-                # print(f"Applying unified pos, x shape: {x.shape}")
                 if self.pos.device != x.device:
                     # move the position tensor:
                     self.pos = self.pos.to(x.device)
@@ -280,23 +304,14 @@ def forward(self, x, fx, T=None):
                     .reshape(x.shape[0], self.H * self.W, self.ref * self.ref)
                     .to(x.device)
                 )
-                # print(f"self.ref is {self.ref}")
-                # print(f"x shape after unified pos: {x.shape}")
             if fx is not None:
-                # print(f"Applying fx, fx shape: {fx.shape}")
                 fx = torch.cat((x, fx), -1)
-                # print(f"fx shape after cat: {fx.shape}")
                 fx = self.preprocess(fx)
-                # print(f"fx shape after preprocess: {fx.shape}")
             else:
-                # print(f"Applying just x, x shape: {x.shape}")
                 fx = self.preprocess(x)
-                # print(f"fx shape after preprocess: {fx.shape}")
                 fx = fx + self.placeholder[None, None, :]
-                # print(f"fx shape after placeholder: {fx.shape}")
 
         if T is not None:
-            # print(f"Applying T, T shape: {T.shape}")
             with torch.autograd.profiler.record_function("Time_Embedding"):
                 Time_emb = timestep_embedding(T, self.n_hidden).repeat(1, x.shape[1], 1)
                 Time_emb = self.time_fc(Time_emb)
@@ -365,6 +380,8 @@ class Transolver(Module):
         The height of the mesh.
     W : int
         The width of the mesh.
+    use_te: bool
+        Whether to use transformer engine backend when possible.
     """
 
     def __init__(
@@ -384,6 +401,7 @@ def __init__(
         unified_pos: bool,
         H: int,
         W: int,
+        use_te: bool = True,
     ) -> None:
         super().__init__(meta=MetaData())
         self.H = H
@@ -404,6 +422,7 @@ def __init__(
             unified_pos=unified_pos,
             H=H,
             W=W,
+            use_te=use_te,
         )
 
     def forward(

From 3d392c379d0eca735ebdead6c2bfdc3e020b79e5 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Tue, 1 Jul 2025 10:01:36 -0700
Subject: [PATCH 03/23] Clean up transolver architecture and code to enable
 irregular and 3D meshes.  Enhance documentation.

---
 examples/cfd/darcy_transolver/config_fix.yaml |  15 +-
 .../train_transolver_darcy_fix.py             |  18 +-
 .../models/transolver/Physics_Attention.py    |  64 +-
 physicsnemo/models/transolver/transolver.py   | 752 +++++++++++++-----
 4 files changed, 608 insertions(+), 241 deletions(-)

diff --git a/examples/cfd/darcy_transolver/config_fix.yaml b/examples/cfd/darcy_transolver/config_fix.yaml
index 7ac13ce133..ce24362bfb 100644
--- a/examples/cfd/darcy_transolver/config_fix.yaml
+++ b/examples/cfd/darcy_transolver/config_fix.yaml
@@ -21,17 +21,17 @@
 # SOFTWARE.
 
 output_dir: ./output/darcy_transolver_fix
-run_id: fp32_dev_test_fullres
+run_id: fp32_dev_fullres_b64_s64-te
 
 data:
   train_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth1.npz
   test_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth2.npz
   resolution: 421 #421, 211, 141, 106, 85 all viable
-  batch_size: 8 # This is the GLOBAL batch size
+  batch_size: 32 # This is the GLOBAL batch size
 
 model:
   space_dim: 2
-  n_layers: 8
+  n_layers: 4
   n_hidden: 128
   dropout: 0.0
   n_head: 8
@@ -40,11 +40,10 @@ model:
   mlp_ratio: 1
   fun_dim: 1
   out_dim: 1
-  # slice_dim: 64
   ref: 8
   unified_pos: False
-  slice_num: 256
-  transformer_engine: False
+  slice_num: 64
+  transformer_engine: True
   
 precision: fp32
 
@@ -58,13 +57,13 @@ normaliser:
 
 scheduler:
   initial_lr: 1.E-3
-  decay_rate: 1.E-5
+  decay_rate: 5.E-5
   weight_decay: 1.E-5
   decay_pseudo_epochs: 8
 
 training:
   rec_results_freq : 100
-  max_pseudo_epochs: 200
+  max_pseudo_epochs: 1000
   pseudo_epoch_sample_size: 1024
 
 validation:
diff --git a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
index e218b67fed..2eea776208 100644
--- a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
+++ b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
@@ -70,6 +70,7 @@ def forward_train_full_loop(
     Forward and backward pass for one iteration, with optional mixed precision training.
     Returns the loss for this minibatch
     """
+    dm = DistributedManager()
     with context:
         pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
         pred = y_normalizer.decode(pred)
@@ -128,19 +129,24 @@ def val_epoch(model, test_dataloader, loss_fun, y_normalizer):
             pred = y_normalizer.decode(pred)
             loss = loss_fun(pred, y)
 
-            # --- Relative L2 calculation ---
-            # print(f"pred shape: {pred.shape}")
-            # print(f"y shape: {y.shape}")
-            rel_l2 = torch.norm(pred.reshape(y.shape) - y) / torch.norm(y)
+            # Compute per-sample relative L2 error
+            diff = pred.reshape(y.shape) - y
+            rel_l2 = torch.norm(diff.view(diff.shape[0], -1), dim=1) / torch.norm(
+                y.view(y.shape[0], -1), dim=1
+            )
+            rel_l2_mean = rel_l2.mean()
+
             if RL2 is None:
-                RL2 = rel_l2
+                RL2 = rel_l2_mean
             else:
-                RL2 += rel_l2
+                RL2 += rel_l2_mean
             if val_loss is None:
                 val_loss = loss
             else:
                 val_loss += loss
+
     val_loss = val_loss / len(test_dataloader)
+    RL2 = RL2 / len(test_dataloader)
 
     dm = DistributedManager()
     if dm.world_size > 1:
diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index 0a169a97f4..c303634391 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -122,21 +122,23 @@ def compute_slices_from_projections(
 
         # Project the latent space vectors on to the weight computation space,
         # and compute a temperature adjusted softmax.
-        clamped_temp = torch.clamp(self.temperature, min=0.1, max=5).to(
+        clamped_temp = torch.clamp(self.temperature, min=0.5, max=5).to(
             slice_projections.dtype
         )
         slice_weights = nn.functional.softmax(
             slice_projections / clamped_temp, dim=-1
         )  # [Batch, N_heads, N_tokens, Slice_num]
+
+        # Cast to the computation type (since the parameter is probably fp32)
         slice_weights = slice_weights.to(slice_projections.dtype)
         # Average the slices over the token dimension
         slice_norm = slice_weights.sum(2)  # [Batch, N_heads, Slice_num]
-
         # This does the projection of the latent space fx by the weights:
-        slice_token = torch.matmul(slice_weights.transpose(2, 3), fx)
 
-        # Apply the normalization (summed weights)
-        slice_token = slice_token / ((slice_norm[:, :, :, None] + 1e-5))  # B H G D
+        # Computing the slice tokens is a matmul followed by a normalization.
+        # It can, unfortunately, overflow in reduced precision, so normalize first:
+        slice_weights = slice_weights / (slice_norm[:, :, None, :] + 1e-2)
+        slice_token = torch.matmul(slice_weights.transpose(2, 3), fx)
 
         return slice_weights, slice_token
 
@@ -226,32 +228,61 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
 
         # Project the inputs onto learned spaces:
         x_mid, fx_mid = self.project_input_onto_slices(x)
+        # if (x_mid.isnan().any()):
+        #     print(f"Rank {dm.rank} x_mid is nan")
+        #     exit()
+        # if (fx_mid.isnan().any()):
+        #     print(f"Rank {dm.rank} fx_mid is nan")
+        #     exit()
         # x_mid and fx_mid should have shapes of [B, N_head, N_tokens, Head_dim]
 
         # Perform the linear projection of learned latent space onto slices:
         slice_projections = self.in_project_slice(x_mid)
-
+        # if (slice_projections.isnan().any()):
+        #     print(f"Rank {dm.rank} slice_projections is nan")
+        #     exit()
         # Slice projections has shape [B, N_head, N_tokens, Head_dim], but head_dim may have changed!
 
         # Use the slice projections and learned spaces to compute the slices, and their weights:
         slice_weights, slice_tokens = self.compute_slices_from_projections(
             slice_projections, fx_mid
         )
-
+        # if (slice_weights.isnan().any()):
+        #     print(f"Rank {dm.rank} slice_weights is nan")
+        #     exit()
+        # if (slice_tokens.isnan().any()) or (slice_tokens.isinf().any()):
+        #     print(f"Rank {dm.rank} slice_tokens is nan or inf")
+        #     torch.save(slice_tokens, f"slice_tokens_{dm.rank}.pt")
+        #     torch.save(slice_weights, f"slice_weights_{dm.rank}.pt")
+        #     torch.save(slice_projections, f"slice_projections_{dm.rank}.pt")
+        #     torch.save(fx_mid, f"fx_mid_{dm.rank}.pt")
+        #     exit()
         # slice_weights has shape [Batch, N_heads, N_tokens, Slice_num]
         # slice_tokens has shape  [Batch, N_heads, N_tokens, head_dim]
 
+        # print(f"Rank {dm.rank} Min/max/std of slice tokens: {slice_tokens.min()}, {slice_tokens.max()}, {slice_tokens.std()}")
+
         # Apply attention to the slice tokens
         if self.use_te:
             out_slice_token = self.compute_slice_attention_te(slice_tokens)
+            # out_slice_token = self.compute_slice_attention(slice_tokens)
         else:
             out_slice_token = self.compute_slice_attention_sdpa(slice_tokens)
 
+        # if (out_slice_token.isnan().any()):
+        #     print(f"Rank {dm.rank} out_slice_token is nan")
+        #     torch.save(slice_tokens, f"slice_tokens_{dm.rank}.pt")
+        #     torch.save(slice_weights, f"slice_weights_{dm.rank}.pt")
+        #     torch.save(slice_projections, f"slice_projections_{dm.rank}.pt")
+        #     torch.save(fx_mid, f"fx_mid_{dm.rank}.pt")
+        #     exit()
         # Shape unchanged
 
         # Deslice:
         outputs = self.project_attention_outputs(out_slice_token, slice_weights)
-
+        # if (outputs.isnan().any()):
+        #     print(f"Rank {dm.rank} outputs is nan")
+        #     exit()
         # Outputs now has the same shape as the original input x
 
         return outputs
@@ -294,20 +325,19 @@ class Physics_Attention_Structured_Mesh_2D_2(Physics_Attention_Base):
     def __init__(
         self,
         dim,
+        spatial_shape,
         heads=8,
         dim_head=64,
         dropout=0.0,
         slice_num=64,
-        H=101,
-        W=31,
         kernel=3,
         use_te=True,
     ):  # kernel=3):
         super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
 
         inner_dim = dim_head * heads
-        self.H = H
-        self.W = W
+        self.H = spatial_shape[0]
+        self.W = spatial_shape[1]
 
         self.in_project_x = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
         self.in_project_fx = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
@@ -347,22 +377,20 @@ class Physics_Attention_Structured_Mesh_3D_2(Physics_Attention_Base):
     def __init__(
         self,
         dim,
+        spatial_shape,
         heads=8,
         dim_head=64,
         dropout=0.0,
         slice_num=32,
-        H=32,
-        W=32,
-        D=32,
         kernel=3,
         use_te=True,
     ):
         super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
 
         inner_dim = dim_head * heads
-        self.H = H
-        self.W = W
-        self.D = D
+        self.H = spatial_shape[0]
+        self.W = spatial_shape[1]
+        self.D = spatial_shape[2]
 
         self.in_project_x = nn.Conv3d(dim, inner_dim, kernel, 1, kernel // 2)
         self.in_project_fx = nn.Conv3d(dim, inner_dim, kernel, 1, kernel // 2)
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index be3f93341f..eaa2c1f49a 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -49,9 +49,15 @@
 from .Embedding import timestep_embedding
 
 # from .Physics_Attention import Physics_Attention_Structured_Mesh_2D
+from .Physics_Attention import (
+    Physics_Attention_Irregular_Mesh_2 as Physics_Attention_Irregular_Mesh,
+)
 from .Physics_Attention import (
     Physics_Attention_Structured_Mesh_2D_2 as Physics_Attention_Structured_Mesh_2D,
 )
+from .Physics_Attention import (
+    Physics_Attention_Structured_Mesh_3D_2 as Physics_Attention_Structured_Mesh_3D,
+)
 
 ACTIVATION = {
     "gelu": nn.GELU,
@@ -99,7 +105,7 @@ def forward(self, x):
 
 
 class Transolver_block(nn.Module):
-    """Transformer encoder block."""
+    """Transformer encoder block, replacing standard attention with physics attention."""
 
     def __init__(
         self,
@@ -111,8 +117,7 @@ def __init__(
         last_layer=False,
         out_dim=1,
         slice_num=32,
-        H=85,
-        W=85,
+        spatial_shape: tuple[int] | None = None,
         use_te=True,
     ):
         super().__init__()
@@ -127,16 +132,41 @@ def __init__(
             self.ln_1 = te.LayerNorm(hidden_dim)
         else:
             self.ln_1 = nn.LayerNorm(hidden_dim)
-        self.Attn = Physics_Attention_Structured_Mesh_2D(
-            hidden_dim,
-            heads=num_heads,
-            dim_head=hidden_dim // num_heads,
-            dropout=dropout,
-            slice_num=slice_num,
-            H=H,
-            W=W,
-            use_te=use_te,
-        )
+
+        if spatial_shape is None:
+            self.Attn = Physics_Attention_Irregular_Mesh(
+                hidden_dim,
+                heads=num_heads,
+                dim_head=hidden_dim // num_heads,
+                dropout=dropout,
+                slice_num=slice_num,
+                use_te=use_te,
+            )
+        else:
+            if len(spatial_shape) == 2:
+                self.Attn = Physics_Attention_Structured_Mesh_2D(
+                    hidden_dim,
+                    spatial_shape=spatial_shape,
+                    heads=num_heads,
+                    dim_head=hidden_dim // num_heads,
+                    dropout=dropout,
+                    slice_num=slice_num,
+                    use_te=use_te,
+                )
+            elif len(spatial_shape) == 3:
+                self.Attn = Physics_Attention_Structured_Mesh_3D(
+                    hidden_dim,
+                    spatial_shape=spatial_shape,
+                    heads=num_heads,
+                    dim_head=hidden_dim // num_heads,
+                    dropout=dropout,
+                    slice_num=slice_num,
+                    use_te=use_te,
+                )
+            else:
+                raise Exception(
+                    f"Unexpected length of spatial shape encountered in Transolver_block: {len(spatial_shape)}"
+                )
 
         if use_te:
             self.ln_mlp1 = te.LayerNormMLP(
@@ -169,67 +199,338 @@ def __init__(
     def forward(self, fx):
         fx = self.Attn(self.ln_1(fx)) + fx
         fx = self.ln_mlp1(fx) + fx
-        # fx = self.mlp(self.ln_2(fx)) + fx
         if self.last_layer:
             return self.ln_mlp2(fx)
-            # return self.mlp2(self.ln_3(fx))
         else:
             return fx
 
 
-class Model(nn.Module):
+# class Model(nn.Module):
+#     def __init__(
+#         self,
+#         space_dim=1,
+#         n_layers=5,
+#         n_hidden=256,
+#         dropout=0.0,
+#         n_head=8,
+#         Time_Input=False,
+#         act="gelu",
+#         mlp_ratio=1,
+#         fun_dim=1,
+#         out_dim=1,
+#         slice_num=32,
+#         ref=8,
+#         unified_pos=False,
+#         H=85,
+#         W=85,
+#         use_te=True,
+#     ):
+#         super().__init__()
+#         self.__name__ = "Transolver_2D"
+#         self.H = H
+#         self.W = W
+#         self.ref = ref
+#         self.unified_pos = unified_pos
+#         if self.unified_pos:
+#             self.pos = self.get_grid()
+#             self.preprocess = MLP(
+#                 fun_dim + self.ref * self.ref,
+#                 n_hidden * 2,
+#                 n_hidden,
+#                 n_layers=0,
+#                 res=False,
+#                 act=act,
+#             )
+#         else:
+#             self.preprocess = MLP(
+#                 fun_dim + space_dim,
+#                 n_hidden * 2,
+#                 n_hidden,
+#                 n_layers=0,
+#                 res=False,
+#                 act=act,
+#             )
+
+#         self.Time_Input = Time_Input
+#         self.n_hidden = n_hidden
+#         self.space_dim = space_dim
+#         if Time_Input:
+#             self.time_fc = nn.Sequential(
+#                 nn.Linear(n_hidden, n_hidden), nn.SiLU(), nn.Linear(n_hidden, n_hidden)
+#             )
+#         self.blocks = nn.ModuleList(
+#             [
+#                 Transolver_block(
+#                     num_heads=n_head,
+#                     hidden_dim=n_hidden,
+#                     dropout=dropout,
+#                     act=act,
+#                     mlp_ratio=mlp_ratio,
+#                     out_dim=out_dim,
+#                     slice_num=slice_num,
+#                     H=H,
+#                     W=W,
+#                     last_layer=(_ == n_layers - 1),
+#                     use_te=use_te,
+#                 )
+#                 for _ in range(n_layers)
+#             ]
+#         )
+#         self.initialize_weights()
+#         # self.placeholder = nn.Parameter(
+#         #     (1 / (n_hidden)) * torch.rand(n_hidden, dtype=torch.float)
+#         # )
+
+#     def initialize_weights(self):
+#         self.apply(self._init_weights)
+
+#     def _init_weights(self, m):
+#         if isinstance(m, nn.Linear):
+#             nn.init.trunc_normal_(m.weight, std=0.02)
+#             if isinstance(m, nn.Linear) and m.bias is not None:
+#                 nn.init.constant_(m.bias, 0)
+#         elif isinstance(m, (nn.LayerNorm, te.LayerNorm, nn.BatchNorm1d)):
+#             nn.init.constant_(m.bias, 0)
+#             nn.init.constant_(m.weight, 1.0)
+
+#     def get_grid(self, batchsize=1):
+#         size_x, size_y = self.H, self.W
+#         gridx = torch.tensor(np.linspace(0, 1, size_x), dtype=torch.float)
+#         gridx = gridx.reshape(1, size_x, 1, 1).repeat([batchsize, 1, size_y, 1])
+#         gridy = torch.tensor(np.linspace(0, 1, size_y), dtype=torch.float)
+#         gridy = gridy.reshape(1, 1, size_y, 1).repeat([batchsize, size_x, 1, 1])
+#         grid = torch.cat((gridx, gridy), dim=-1)  # B H W 2
+
+#         gridx = torch.tensor(np.linspace(0, 1, self.ref), dtype=torch.float)
+#         gridx = gridx.reshape(1, self.ref, 1, 1).repeat([batchsize, 1, self.ref, 1])
+#         gridy = torch.tensor(np.linspace(0, 1, self.ref), dtype=torch.float)
+#         gridy = gridy.reshape(1, 1, self.ref, 1).repeat([batchsize, self.ref, 1, 1])
+#         grid_ref = torch.cat((gridx, gridy), dim=-1)  # B H W 8 8 2
+
+#         pos = (
+#             torch.sqrt(
+#                 torch.sum(
+#                     (grid[:, :, :, None, None, :] - grid_ref[:, None, None, :, :, :])
+#                     ** 2,
+#                     dim=-1,
+#                 )
+#             )
+#             .reshape(batchsize, size_x, size_y, self.ref * self.ref)
+#             .contiguous()
+#         )
+#         return pos
+
+#     def forward(self, x, fx, T=None):
+#         with torch.autograd.profiler.record_function("prepeocess"):
+#             if self.unified_pos:
+#                 if self.pos.device != x.device:
+#                     # move the position tensor:
+#                     self.pos = self.pos.to(x.device)
+
+#                 x = (
+#                     self.pos.repeat(x.shape[0], 1, 1, 1)
+#                     .reshape(x.shape[0], self.H * self.W, self.ref * self.ref)
+#                     .to(x.device)
+#                 )
+#     if fx is not None:
+#         fx = torch.cat((x, fx), -1)
+#         fx = self.preprocess(fx)
+#     else:
+#         fx = self.preprocess(x)
+#         fx = fx + self.placeholder[None, None, :]
+
+# if T is not None:
+#     with torch.autograd.profiler.record_function("Time_Embedding"):
+#         Time_emb = timestep_embedding(T, self.n_hidden).repeat(1, x.shape[1], 1)
+#         Time_emb = self.time_fc(Time_emb)
+#         fx = fx + Time_emb
+
+# for i, block in enumerate(self.blocks):
+#     with torch.autograd.profiler.record_function(f"Block_{i}"):
+#         fx = block(fx)
+
+# return fx
+
+
+@dataclass
+class MetaData(ModelMetaData):
+    name: str = "Transolver"
+    # Optimization
+    jit: bool = False
+    cuda_graphs: bool = False
+    amp: bool = False
+    # Inference
+    onnx_cpu: bool = False  # No FFT op on CPU
+    onnx_gpu: bool = True
+    onnx_runtime: bool = True
+    # Physics informed
+    var_dim: int = 1
+    func_torch: bool = False
+    auto_grad: bool = False
+
+
+class Transolver(Module):
+    """
+    Transolver model, adapted from original transolver code.
+
+    Transolver is an adaptation of the transformer architecture, with a physics-attention
+    mechanism replacing the standard attention mechanism.
+
+    For more architecture details, see: https://arxiv.org/pdf/2402.02366 and https://arxiv.org/pdf/2502.02414
+
+    Transolver can work on structured or unstructured data points as a model construction choice:
+    - unstructured data (like a mesh) should provide some sort of positional encoding to accompany inputs
+    - structured data (2D and 3D grids) can provide positional encodings optionally
+
+    When constructing Transolver, you can choose to use "unified position" or not.  If you select "unified
+    position" (`unified_pos=True`), then
+
+    If using structured data, pass the structured shape as a tuple in the model constructor.
+    Length 2 tuples are assumed to be image-like, length 3 tuples are assumed to be 3D voxel like.
+    Other structured shape sizes are not supported.  Passing a structured_shape of None assumes irregular data.
+
+    Output shape will have the same spatial shape as the input shape, with potentially more features
+
+    Also can support Transolver++ implementation.  When using the distributed algorithm
+    of Transolver++, use PhysicsNeMo's ShardTensor implementation to support automatic
+    domain parallelism and 2D parallelization (data parallel + domain parallel, for example).
+
+    Note
+    ----
+
+
+    Parameters
+    ----------
+    functional_dim : int
+        The dimension of the input values, not including any embeddings.  No Default.
+        Input will be concatenated with embeddings or unified position before processing
+        with PhysicsAttention blocks.  Originally known as "fun_dim"
+    out_dim : int
+        The dimension of the output of the model.  This is a mandatory parameter.
+    embedding_dim : int | None
+        The spatial dimension of the input data embeddings.  Should include not just
+        position but all computed embedding features.  Default is None, but if
+        `unified_pos=False` this is a mandatory parameter.  Originally named "space_dim"
+    n_layers : int
+        The number of transformer PhysicsAttention layers in the model.  Default of 4.
+    n_hidden : int
+        The hidden dimension of the transformer.  Default of 256.  Projection is made
+        from the input data + embeddings in the early preprocessing, before the
+        PhysicsAttention layers.
+    dropout : float
+        The dropout rate, applied across the PhysicsAttention Layers.  Default is 0.0
+    n_head : int
+        The number of attention heads in each PhysicsAttention Layer.  Default is 8.  Note
+        that the number of heads must evenly divide the `n_hidden` parameter to yield an
+        integer head dimension.
+    act : str
+        The activation function, default is Gelu.
+    mlp_ratio : int
+        The ratio of hidden dimension in the MLP, default is 4.  Used in the MLPs in the
+        PhysicsAttention Layers.
+    slice_num : int
+        The number of slices in the PhysicsAttention layers.  Default is 32.  Represents the
+        number of learned states each layer should project inputs onto.
+    unified_pos : bool
+        Whether to use unified positional embeddings.  Unified positions are only available for
+        structured data (2D grids, 3D grids).  They are computed once initially, and reused through
+        training in place of embeddings.
+    ref : int
+        The reference dimension size when using unified positions.  Default is 8.  Will be
+        used to create a linear grid in spatial dimensions to serve as spatial embeddings.
+        If `unified_pos=False`, this value is unused.
+    structured_shape : None | tuple(int)
+        The shape of the latent space.  If None, assumes irregular latent space.  If not
+        `None`, this parameter can only be a length-2 or length-3 tuple of ints.
+    use_te: bool
+        Whether to use transformer engine backend when possible.
+    Time_Input : bool
+        Whether to include time embeddings. Default is false
+    """
+
     def __init__(
         self,
-        space_dim=1,
-        n_layers=5,
-        n_hidden=256,
-        dropout=0.0,
-        n_head=8,
-        Time_Input=False,
-        act="gelu",
-        mlp_ratio=1,
-        fun_dim=1,
-        out_dim=1,
-        slice_num=32,
-        ref=8,
-        unified_pos=False,
-        H=85,
-        W=85,
-        use_te=True,
-    ):
-        super().__init__()
-        self.__name__ = "Transolver_2D"
-        self.H = H
-        self.W = W
-        self.ref = ref
-        self.unified_pos = unified_pos
-        if self.unified_pos:
-            self.pos = self.get_grid()
-            self.preprocess = MLP(
-                fun_dim + self.ref * self.ref,
-                n_hidden * 2,
-                n_hidden,
-                n_layers=0,
-                res=False,
-                act=act,
+        functional_dim: int,
+        out_dim: int,
+        embedding_dim: int | None = None,
+        n_layers: int = 4,
+        n_hidden: int = 256,
+        dropout: float = 0.0,
+        n_head: int = 8,
+        act: str = "gelu",
+        mlp_ratio: int = 4,
+        slice_num: int = 32,
+        unified_pos: bool = False,
+        ref: int = 8,
+        structured_shape: None | tuple[int] = None,
+        use_te: bool = True,
+        Time_Input: bool = False,
+    ) -> None:
+        super().__init__(meta=MetaData())
+        self.__name__ = "Transolver"
+
+        # Check that the hidden dimension and head dimensions are compatible:
+        if not n_hidden % n_head == 0:
+            raise ValueError(
+                f"Transolver requires n_hidden % n_head == 0, but instead got {n_hidden % n_head}"
             )
+
+        # Check the shape of the data, if it's structured data:
+        if structured_shape is not None:
+            # Has to be 2D or 3D data:
+            if len(structured_shape) not in [2, 3]:
+                raise ValueError(
+                    f"Transolver can only use structured data in 2D or 3D, got {structured_shape}"
+                )
+
+            # Ensure it's all integers > 0:
+            if not all([s > 0 and s == int(s) for s in structured_shape]):
+                raise ValueError(
+                    f"Transolver can only use integer shapes > 0, got {structured_shape}"
+                )
         else:
-            self.preprocess = MLP(
-                fun_dim + space_dim,
-                n_hidden * 2,
-                n_hidden,
-                n_layers=0,
-                res=False,
-                act=act,
-            )
+            # It's mandatory for unified position:
+            if unified_pos:
+                raise ValueError(
+                    "Transolver requires structured_shape to be passed if using unified_pos=True"
+                )
+
+        self.structured_shape = structured_shape
+
+        # If we're using the unified position, create and save the position embeddings:
+
+        if unified_pos:
+            if structured_shape is None:
+                raise ValueError(
+                    "Transolver can not use unified position without a structured_shape argument (got None)"
+                )
+
+            # This ensures embedding is tracked by torch and moves to the GPU, and saves/loads
+            self.register_buffer("embedding", self.get_grid(ref))
+            self.embedding_dim = ref * ref
+            mlp_input_dimension = functional_dim + ref * ref
+
+        else:
+            self.embedding_dim = embedding_dim
+            mlp_input_dimension = functional_dim + embedding_dim
+
+        # This MLP is the initial projection onto the hidden space
+        self.preprocess = MLP(
+            mlp_input_dimension,
+            n_hidden * 2,
+            n_hidden,
+            n_layers=0,
+            res=False,
+            act=act,
+        )
 
         self.Time_Input = Time_Input
         self.n_hidden = n_hidden
-        self.space_dim = space_dim
         if Time_Input:
             self.time_fc = nn.Sequential(
                 nn.Linear(n_hidden, n_hidden), nn.SiLU(), nn.Linear(n_hidden, n_hidden)
             )
+
         self.blocks = nn.ModuleList(
             [
                 Transolver_block(
@@ -240,8 +541,7 @@ def __init__(
                     mlp_ratio=mlp_ratio,
                     out_dim=out_dim,
                     slice_num=slice_num,
-                    H=H,
-                    W=W,
+                    spatial_shape=structured_shape,
                     last_layer=(_ == n_layers - 1),
                     use_te=use_te,
                 )
@@ -265,18 +565,34 @@ def _init_weights(self, m):
             nn.init.constant_(m.bias, 0)
             nn.init.constant_(m.weight, 1.0)
 
-    def get_grid(self, batchsize=1):
-        size_x, size_y = self.H, self.W
+    def get_grid(self, ref: int, batchsize: int = 1) -> torch.Tensor:
+        """
+        Generate a unified positional encoding grid for structured 2D data.
+
+        Parameters
+        ----------
+        ref : int
+            The reference grid size for the unified position encoding.
+        batchsize : int, optional
+            The batch size for the generated grid (default is 1).
+
+        Returns
+        -------
+        torch.Tensor
+            A tensor of shape (batchsize, H*W, ref*ref) containing the positional encodings,
+            where H and W are the spatial dimensions from self.structured_shape.
+        """
+        size_x, size_y = self.structured_shape
         gridx = torch.tensor(np.linspace(0, 1, size_x), dtype=torch.float)
         gridx = gridx.reshape(1, size_x, 1, 1).repeat([batchsize, 1, size_y, 1])
         gridy = torch.tensor(np.linspace(0, 1, size_y), dtype=torch.float)
         gridy = gridy.reshape(1, 1, size_y, 1).repeat([batchsize, size_x, 1, 1])
         grid = torch.cat((gridx, gridy), dim=-1)  # B H W 2
 
-        gridx = torch.tensor(np.linspace(0, 1, self.ref), dtype=torch.float)
-        gridx = gridx.reshape(1, self.ref, 1, 1).repeat([batchsize, 1, self.ref, 1])
-        gridy = torch.tensor(np.linspace(0, 1, self.ref), dtype=torch.float)
-        gridy = gridy.reshape(1, 1, self.ref, 1).repeat([batchsize, self.ref, 1, 1])
+        gridx = torch.tensor(np.linspace(0, 1, ref), dtype=torch.float)
+        gridx = gridx.reshape(1, ref, 1, 1).repeat([batchsize, 1, ref, 1])
+        gridy = torch.tensor(np.linspace(0, 1, ref), dtype=torch.float)
+        gridy = gridy.reshape(1, 1, ref, 1).repeat([batchsize, ref, 1, 1])
         grid_ref = torch.cat((gridx, gridy), dim=-1)  # B H W 8 8 2
 
         pos = (
@@ -287,33 +603,66 @@ def get_grid(self, batchsize=1):
                     dim=-1,
                 )
             )
-            .reshape(batchsize, size_x, size_y, self.ref * self.ref)
+            .reshape(batchsize, -1, ref * ref)  # Flatten spatial dims
             .contiguous()
         )
         return pos
 
-    def forward(self, x, fx, T=None):
+    def forward(
+        self,
+        fx: torch.Tensor,
+        embedding: torch.Tensor | None = None,
+        T: torch.Tensor = None,
+    ) -> torch.Tensor:
+        """
+        Forward pass of the transolver model.
+
+        For structured data, you may pass the functional input and embedding
+        as either [B, N, C] shape or [B, *structure, C]. Output will be returned
+        in the same shape as input.
+
+        Unstructured data must be passed as [B, N, C] for both embeddings and tokens.
+
+        Sublayers expect shape of [B, N, C]
+
+        """
         with torch.autograd.profiler.record_function("prepeocess"):
+
+            # Reshape automatically, if necessary:
+            if self.structured_shape is not None:
+                unflatten_output = False
+                if len(fx.shape) != 3:
+                    unflatten_output = True
+                    fx = fx.reshape(fx.shape[0], -1, fx.shape[-1])
+                if embedding is not None and len(embedding.shape) != 3:
+                    embedding = embedding.reshape(
+                        embedding.shape[0], *self.structured_shape, -1
+                    )
+
             if self.unified_pos:
-                if self.pos.device != x.device:
-                    # move the position tensor:
-                    self.pos = self.pos.to(x.device)
-
-                x = (
-                    self.pos.repeat(x.shape[0], 1, 1, 1)
-                    .reshape(x.shape[0], self.H * self.W, self.ref * self.ref)
-                    .to(x.device)
+                # Extend the embedding to the batch size:
+                embedding = (
+                    self.embedding.repeat(embedding.shape[0], 1, 1)
+                    # .reshape(x.shape[0], -1, self.embedding_dim)
                 )
-            if fx is not None:
-                fx = torch.cat((x, fx), -1)
-                fx = self.preprocess(fx)
-            else:
-                fx = self.preprocess(x)
-                fx = fx + self.placeholder[None, None, :]
+
+            # Combine the embedding and functional input:
+            fx = torch.cat((embedding, fx), -1)
+
+            # Dead code?  Not sure when this is used.
+            # if fx is not None:
+            # else:
+            #     fx = self.preprocess(x)
+            #     fx = fx + self.placeholder[None, None, :]
+
+            # Apply preprocessing
+            fx = self.preprocess(fx)
 
         if T is not None:
             with torch.autograd.profiler.record_function("Time_Embedding"):
-                Time_emb = timestep_embedding(T, self.n_hidden).repeat(1, x.shape[1], 1)
+                Time_emb = timestep_embedding(T, self.n_hidden).repeat(
+                    1, embedding.shape[1], 1
+                )
                 Time_emb = self.time_fc(Time_emb)
                 fx = fx + Time_emb
 
@@ -321,133 +670,118 @@ def forward(self, x, fx, T=None):
             with torch.autograd.profiler.record_function(f"Block_{i}"):
                 fx = block(fx)
 
-        return fx
-
+        if self.structured_shape is not None:
+            if unflatten_output:
+                fx = fx.reshape(fx.shape[0], *self.structured_shape, -1)
 
-@dataclass
-class MetaData(ModelMetaData):
-    name: str = "Transolver"
-    # Optimization
-    jit: bool = False
-    cuda_graphs: bool = False
-    amp: bool = False
-    # Inference
-    onnx_cpu: bool = False  # No FFT op on CPU
-    onnx_gpu: bool = True
-    onnx_runtime: bool = True
-    # Physics informed
-    var_dim: int = 1
-    func_torch: bool = False
-    auto_grad: bool = False
-
-
-class Transolver(Module):
-    """Transformer-based solver for PDEs.
-
-    Note
-    ----
-    Transolver is a model specifically designed for structured 2D mesh data.
-
-    Parameters
-    ----------
-    space_dim : int
-        The spatial dimension of the input data.
-    n_layers : int
-        The number of transformer layers.
-    n_hidden : int
-        The hidden dimension of the transformer.
-    dropout : float
-        The dropout rate.
-    n_head : int
-        The number of attention heads.
-    Time_Input : bool
-        Whether to include time embeddings.
-    act : str
-        The activation function.
-    mlp_ratio : int
-        The ratio of hidden dimension in the MLP.
-    fun_dim : int
-        The dimension of the function.
-    out_dim : int
-        The output dimension.
-    slice_num : int
-        The number of slices in the structured attention.
-    ref : int
-        The reference dimension.
-    unified_pos : bool
-        Whether to use unified positional embeddings.
-    H : int
-        The height of the mesh.
-    W : int
-        The width of the mesh.
-    use_te: bool
-        Whether to use transformer engine backend when possible.
-    """
-
-    def __init__(
-        self,
-        space_dim: int,
-        n_layers: int,
-        n_hidden: int,
-        dropout: float,
-        n_head: int,
-        Time_Input: bool,
-        act: str,
-        mlp_ratio: int,
-        fun_dim: int,
-        out_dim: int,
-        slice_num: int,
-        ref: int,
-        unified_pos: bool,
-        H: int,
-        W: int,
-        use_te: bool = True,
-    ) -> None:
-        super().__init__(meta=MetaData())
-        self.H = H
-        self.W = W
-        self.model = Model(
-            space_dim=space_dim,
-            n_layers=n_layers,
-            n_hidden=n_hidden,
-            dropout=dropout,
-            n_head=n_head,
-            Time_Input=Time_Input,
-            act=act,
-            mlp_ratio=mlp_ratio,
-            fun_dim=fun_dim,
-            out_dim=out_dim,
-            slice_num=slice_num,
-            ref=ref,
-            unified_pos=unified_pos,
-            H=H,
-            W=W,
-            use_te=use_te,
-        )
-
-    def forward(
-        self, x: torch.Tensor, fx: torch.Tensor = None, T: torch.Tensor = None
-    ) -> torch.Tensor:
-        """Forward pass.
-
-        Parameters
-        ----------
-        x : torch.Tensor
-            The input tensor.
-        fx : torch.Tensor
-            The function tensor.
-        T : torch.Tensor
-            The time tensor.
+        return fx
 
-        Returns
-        -------
-        torch.Tensor
-            The output tensor.
 
-        """
-        # print(f"Input x shape: {x.shape}")
-        # print(f"Input fx shape: {fx.shape if fx is not None else None}")
-        # print(f"Input T shape: {T.shape if T is not None else None}")
-        y = self.model(x, fx, T)
-        y = y.reshape(x.shape[0], self.H, self.W, -1)
-        return y
+# class Transolver(Module):
+#     """Transformer-based solver for PDEs.
+
+#     Note
+#     ----
+#     Transolver is a model specifically designed for structured 2D mesh data.
+
+#     Parameters
+#     ----------
+#     space_dim : int
+#         The spatial dimension of the input data.
+#     n_layers : int
+#         The number of transformer layers.
+#     n_hidden : int
+#         The hidden dimension of the transformer.
+#     dropout : float
+#         The dropout rate.
+#     n_head : int
+#         The number of attention heads.
+#     Time_Input : bool
+#         Whether to include time embeddings.
+#     act : str
+#         The activation function.
+#     mlp_ratio : int
+#         The ratio of hidden dimension in the MLP.
+#     fun_dim : int
+#         The dimension of the function.
+#     out_dim : int
+#         The output dimension.
+#     slice_num : int
+#         The number of slices in the structured attention.
+#     ref : int
+#         The reference dimension.
+#     unified_pos : bool
+#         Whether to use unified positional embeddings.
+#     structured_shape : None | tuple(int)
+#         The shape of the latent space.  If None, assumes irregulare latent space.
+#     use_te: bool
+#         Whether to use transformer engine backend when possible.
+#     """
+
+#     def __init__(
+#         self,
+#         space_dim: int,
+#         n_layers: int,
+#         n_hidden: int,
+#         dropout: float,
+#         n_head: int,
+#         Time_Input: bool,
+#         act: str,
+#         mlp_ratio: int,
+#         fun_dim: int,
+#         out_dim: int,
+#         slice_num: int,
+#         ref: int,
+#         unified_pos: bool,
+#         H: int,
+#         W: int,
+#         use_te: bool = True,
+#     ) -> None:
+#         super().__init__(meta=MetaData())
+#         self.H = H
+#         self.W = W
+#         self.model = Model(
+#             space_dim=space_dim,
+#             n_layers=n_layers,
+#             n_hidden=n_hidden,
+#             dropout=dropout,
+#             n_head=n_head,
+#             Time_Input=Time_Input,
+#             act=act,
+#             mlp_ratio=mlp_ratio,
+#             fun_dim=fun_dim,
+#             out_dim=out_dim,
+#             slice_num=slice_num,
+#             ref=ref,
+#             unified_pos=unified_pos,
+#             H=H,
+#             W=W,
+#             use_te=use_te,
+#         )
+
+#     def forward(
+#         self, x: torch.Tensor, fx: torch.Tensor = None, T: torch.Tensor = None
+#     ) -> torch.Tensor:
+#         """Forward pass.
+
+#         Parameters
+#         ----------
+#         x : torch.Tensor
+#             The input tensor.
+#         fx : torch.Tensor
+#             The function tensor.
+#         T : torch.Tensor
+#             The time tensor.
+
+#         Returns
+#         -------
+#         torch.Tensor
+#             The output tensor.
+
+#         """
+#         # print(f"Input x shape: {x.shape}")
+#         # print(f"Input fx shape: {fx.shape if fx is not None else None}")
+#         # print(f"Input T shape: {T.shape if T is not None else None}")
+#         y = self.model(x, fx, T)
+#         y = y.reshape(x.shape[0], self.H, self.W, -1)
+#         return y

From c0517e567c5cae15097943443714ab86983cb34e Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Wed, 2 Jul 2025 15:50:00 +0000
Subject: [PATCH 04/23] Beginning work on transolver example.  Planning ahead
 for domain parallelism, this is a disk to gpu pipeline that natively
 optimizes bandwidth when domain parallel.

---
 .../transolver/README.md                      |   0
 .../transolver/datapipe.py                    | 510 ++++++++++++++++++
 2 files changed, 510 insertions(+)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/README.md
 create mode 100644 examples/cfd/external_aerodynamics/transolver/datapipe.py

diff --git a/examples/cfd/external_aerodynamics/transolver/README.md b/examples/cfd/external_aerodynamics/transolver/README.md
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/examples/cfd/external_aerodynamics/transolver/datapipe.py b/examples/cfd/external_aerodynamics/transolver/datapipe.py
new file mode 100644
index 0000000000..c1de86b431
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/datapipe.py
@@ -0,0 +1,510 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: MIT License
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+"""
+Domain-parallel Zarr dataset for Transolver CFD data.
+
+This dataset implements domain parallelism where multiple ranks can read chunks
+of data cooperatively from zarr files. It supports conversion to ShardTensor
+objects for distributed training.
+"""
+
+import os
+import time
+from concurrent.futures import ThreadPoolExecutor, Future
+from pathlib import Path
+from typing import Dict, List, Optional, Tuple, Union
+from functools import lru_cache
+
+import numpy as np
+import zarr
+
+import torch
+import torch.distributed as dist
+from torch.distributed.device_mesh import DeviceMesh
+from torch.distributed.tensor.placement_types import Placement, Replicate, Shard
+from torch.distributed.tensor._dtensor_spec import TensorMeta
+from torch.utils.data import Dataset
+
+from physicsnemo.distributed import DistributedManager
+from physicsnemo.distributed.shard_tensor import ShardTensor
+from physicsnemo.distributed._shard_tensor_spec import (
+    ShardTensorSpec,
+    _stride_from_contiguous_shape_C_style,
+)
+from physicsnemo.distributed.utils import compute_split_shapes
+from physicsnemo.utils.profiling import profile
+
+
+def get_filenames(data_path: Path, exclude_dirs: bool = True) -> List[str]:
+    """Get list of filenames from data directory."""
+
+    filenames = []
+    for item in data_path.iterdir():
+        # if exclude_dirs and item.is_dir():
+        #     continue
+        if item.suffix in [".zarr"]:
+            filenames.append(item.name)
+    return sorted(filenames)
+
+
+def _read_chunk_into_array(
+    cpu_array: np.ndarray,
+    zarr_array: zarr.Array,
+    cpu_slice: slice,
+    zarr_slice: slice = None,
+):
+    """Helper function to read a chunk from zarr into numpy array."""
+    if zarr_slice is None:
+        zarr_slice = cpu_slice
+    cpu_array[cpu_slice] = zarr_array[zarr_slice]
+
+
+@lru_cache
+def to_torch_dtype(dtype: np.dtype) -> torch.dtype:
+    """Convert a numpy dtype to a torch dtype."""
+    temp = torch.from_numpy(np.empty((), dtype=dtype))
+    return temp.dtype
+
+
+class DomainParallelZarrDataset(Dataset):
+    """
+    PyTorch dataset for domain-parallel reading of Zarr files.
+
+    This dataset supports:
+    - Efficient chunk-aligned reading for large data
+    - GPU memory streaming (disk -> numpy -> torch.Tensor -> GPU)
+    - Domain parallelism via DeviceMesh configuration
+    - Conversion to ShardTensor objects for domain-parallel usage
+
+    This dataset supports multiple processes cooperating to read the data, and optimizations
+    for single process reading.
+    - The dataset will only read the keys from the `keys_to_read` parameter.
+    - For keys in `large_keys`, the dataset will spawn threads to read the data,
+      where each thread is aligned with a zarr chunk whenever possible.
+    - If domain parallelism is enabled, then the large and non-large keys may have alterate behavior:
+      - For large keys, each rank will participate in the read.  Each rank will read a portion of the data,
+        designed to be approximately equal portions per process.  Depending on the final output placement
+        of that key, it will be allgathered on the GPU (replicated) or converted inplace to ShardTensor.
+      - For small keys, each rank will read each key in its entirety.  If the output placement is sharded,
+        the keys will be converted in place.  A future update may enable round-robin reading, where each
+        process will read only a selection of keys and broadcast to partners.
+
+    Technical details:
+    - CPU memory is allocated with torch, and if pin_memory is True, it's allocated directly into
+      pinned memory space.  Numpy wraps the torch allocated memory, and Zarr streams the data into it.
+    - The threading is done on the zarr read only - typically the main bottleneck is the memory allocation
+      with pytorch.
+      - There might be a benefit to using threads on the torch alloc - but would require mutliple levels of threading.
+
+
+    Args:
+        data_path: Path to directory containing zarr files
+        device_mesh: Optional DeviceMesh for domain parallelism. If None, reads full data locally.
+        placements: Placement specifications for sharding. Must match mesh dimensions.
+        max_workers: Maximum number of threads for parallel I/O
+        pin_memory: Whether to pin the CPU memory of the allocated tensors.  If True, the memory is allocated
+                    in pinned memory space.  If False, the memory is allocated in the default memory space.
+                    Pinning is faster for GPU transfers, but the available pinned memory is smaller.
+        keys_to_read: Specific keys to read from zarr files. If None, reads all standard keys.
+        large_keys: Keys that should use chunk-aligned reading (example: high res volume data)
+    """
+
+    def __init__(
+        self,
+        data_path: Union[str, Path],
+        device_mesh: DeviceMesh | None = None,
+        placements: Placement | dict[str, Placement] | None = None,
+        max_workers: int = 4,
+        pin_memory: bool = True,
+        keys_to_read: list[str] | None = None,
+        large_keys: set[str] | None = None,
+    ):
+        super().__init__()
+
+        # Initialize distributed manager if not already done
+        if not DistributedManager.is_initialized():
+            DistributedManager.initialize()
+
+        self.dm = DistributedManager()
+        self.data_path = Path(data_path) if isinstance(data_path, str) else data_path
+        self.device_mesh = device_mesh
+        self.placements = placements
+        self.max_workers = max_workers
+        self.pin_memory = pin_memory
+        self.keys_to_read = keys_to_read or set()
+        self.large_keys = large_keys or set()
+
+        # Validate distributed configuration
+        if self.device_mesh is not None:
+            if self.device_mesh.ndim != 1:
+                raise ValueError(
+                    "DomainParallelZarrDataset requires a single axis DeviceMesh if used"
+                )
+            if self.placements is None:
+                raise ValueError(
+                    "placements must be specified when device_mesh is provided"
+                )
+            if isinstance(self.placements, dict):
+                for key, placement in self.placements.items():
+                    if len(placement) != 1:
+                        raise ValueError(
+                            f"placements must be a single placement for each key, got {placement} for key {key}"
+                        )
+                # Check that each key to read has a placement:
+                for key in self.keys_to_read:
+                    if key not in self.placements:
+                        raise ValueError(
+                            f"placements must specify a placement for each key to read, missing placement for key {key}"
+                        )
+            elif isinstance(self.placements, tuple):
+                if len(self.placements) != 1:
+                    raise ValueError(
+                        f"placements must be a length-1 tuple of placement if not a dict, got {self.placements}"
+                    )
+            else:
+                raise ValueError(
+                    f"placements must be a dict or tuple, got {type(self.placements)}"
+                )
+
+        # Get list of zarr files
+        self.filenames = get_filenames(self.data_path, exclude_dirs=True)
+        if not self.filenames:
+            raise ValueError(f"No zarr files found in {self.data_path}")
+
+        # Create thread pool for parallel I/O
+        self.executor = ThreadPoolExecutor(max_workers=self.max_workers)
+
+        # We cache ShardTensorSpecs for each tensor read, based on how they are
+        # read and not how they are meant to be.  They get converted in the end.
+        self.tensor_specs = {}
+
+    def __len__(self) -> int:
+        return len(self.filenames)
+
+    def __del__(self):
+        """Clean up thread pool on destruction."""
+        if hasattr(self, "executor"):
+            self.executor.shutdown(wait=True)
+
+    @profile
+    def _read_key_chunk_aligned(
+        self, zarr_group: zarr.Group, key: str, futures: list[Future]
+    ) -> np.ndarray:
+        """
+        Read a key using chunk-aligned I/O for efficiency.
+        Implements domain parallelism if device_mesh is configured.
+        """
+
+        zarr_array = zarr_group[key]
+
+        # Determine what slice this rank should read
+        if self.device_mesh is not None:
+            # How many splits to make?
+            n_splits = dist.get_world_size(group=self.device_mesh.get_group())
+            # What rank is this one?
+            this_rank = self.device_mesh.get_local_rank()
+
+            sections = compute_split_shapes(zarr_array.shape[0], n_splits)
+
+            global_chunk_start = sum(sections[:this_rank])
+            global_chunk_stop = global_chunk_start + sections[this_rank]
+
+            chunk_sizes = tuple(
+                (section,) + zarr_array.shape[1:] for section in sections
+            )
+            stride = _stride_from_contiguous_shape_C_style(zarr_array.shape)
+
+            meta = TensorMeta(zarr_array.shape, stride, zarr_array.dtype)
+            self.tensor_specs[key] = ShardTensorSpec(
+                mesh=self.device_mesh,
+                placements=(Shard(0),),
+                tensor_meta=meta,
+                _sharding_shapes={0: chunk_sizes},
+            )
+
+        else:
+            global_chunk_start, global_chunk_stop = 0, zarr_array.shape[0]
+
+        # Calculate the shape of data this rank will read
+        local_shape = [global_chunk_stop - global_chunk_start] + list(
+            zarr_array.shape[1:]
+        )
+
+        # Pre-allocate result array
+        torch_output = torch.empty(
+            local_shape,
+            dtype=to_torch_dtype(zarr_array.dtype),
+            pin_memory=self.pin_memory,
+        )
+        # Share the memory buffer with numpy:
+        result = torch_output.numpy()
+
+        # For chunk-aligned reading, align with zarr's chunk boundaries
+        # This is easiest when we precompute start/end slices
+        zarr_chunk_size = zarr_array.chunks[0]
+
+        # Generate the global list of chunk boundaries first (then apply corrections)
+        slice_starts = list(range(0, zarr_array.shape[0], zarr_chunk_size))
+        slice_stops = [start + zarr_chunk_size for start in slice_starts]
+
+        # Correct the last stop:
+        slice_stops[-1] = zarr_array.shape[0]
+
+        # Now, select all the slices that this rank is responsible for:
+        # These are all the boundary points exclusively within the global chunk slice:
+        local_boundaries = [
+            s for s in slice_starts if s >= global_chunk_start and s < global_chunk_stop
+        ]
+
+        # Fix the boundaries:
+        if global_chunk_start not in local_boundaries:
+            zarr_slice_starts = [global_chunk_start] + local_boundaries
+        else:
+            zarr_slice_starts = [s for s in local_boundaries]
+
+        # The stops are the +1 locations
+        zarr_slice_stops = [s for s in zarr_slice_starts[1:]]
+
+        # Handle the end situation
+        if global_chunk_stop not in zarr_slice_stops:
+            zarr_slice_stops.append(global_chunk_stop)
+        else:
+            # It's already in the list of boundaries
+            # We have to reduce the list of starts:
+            zarr_slice_starts = zarr_slice_starts[:-1]
+
+        slice_sizes = [
+            stop - start for stop, start in zip(zarr_slice_stops, zarr_slice_starts)
+        ]
+
+        cpu_slice_starts = [0]
+        cpu_slice_stops = [slice_sizes[0]]
+        for slice_size in slice_sizes[1:]:
+            cpu_slice_starts.append(cpu_slice_stops[-1])
+            cpu_slice_stops.append(cpu_slice_starts[-1] + slice_size)
+
+        # Now, spawn threads to do each of those reads.
+
+        for i in range(len(slice_sizes)):
+
+            zarr_slice = np.s_[zarr_slice_starts[i] : zarr_slice_stops[i]]
+            cpu_slice = np.s_[cpu_slice_starts[i] : cpu_slice_stops[i]]
+
+            future = self.executor.submit(
+                _read_chunk_into_array,
+                result,
+                zarr_array,
+                cpu_slice,
+                zarr_slice,
+            )
+            futures.append(future)
+
+        return torch_output
+
+    @profile
+    def _read_key_standard(
+        self, zarr_group: zarr.Group, key: str, futures: list[Future]
+    ) -> torch.Tensor:
+        """Read a key with simple I/O (for small arrays)."""
+        zarr_array = zarr_group[key]
+
+        # Handle scalar values
+        if zarr_array.shape == ():
+
+            if self.device_mesh is not None:
+                self.tensor_specs[key] = ShardTensorSpec(
+                    mesh=self.device_mesh,
+                    placements=(Replicate(),),
+                    tensor_meta=TensorMeta(zarr_array.shape, (), zarr_array.dtype),
+                    _sharding_shapes={},
+                )
+
+            output = torch.from_numpy(np.array(zarr_array))
+            if self.pin_memory:
+                output = output.pin_memory()
+            return output
+
+        # data = np.empty(zarr_array.shape, dtype=zarr_array.dtype)
+        output = torch.empty(
+            zarr_array.shape,
+            dtype=to_torch_dtype(zarr_array.dtype),
+            pin_memory=self.pin_memory,
+        )
+        data = output.numpy()
+        slice = np.s_[:]
+        futures.append(
+            self.executor.submit(
+                _read_chunk_into_array,
+                data,
+                zarr_array,
+                slice,
+            )
+        )
+
+        if self.device_mesh is not None:
+            # We need to track the tensor meta for this key
+            stride = _stride_from_contiguous_shape_C_style(zarr_array.shape)
+
+            self.tensor_specs[key] = ShardTensorSpec(
+                mesh=self.device_mesh,
+                placements=(Replicate(),),
+                tensor_meta=TensorMeta(zarr_array.shape, stride, zarr_array.dtype),
+                _sharding_shapes={},
+            )
+
+        return output
+
+    @profile
+    def _read_zarr_file(self, filepath: Path) -> Dict[str, np.ndarray]:
+        """Read data from a zarr file."""
+        with zarr.open_group(filepath, mode="r") as zarr_group:
+            data = {}
+            futures = []
+
+            # Process each key
+            for key in self.keys_to_read:
+
+                if key not in zarr_group:
+                    continue
+
+                if key in self.large_keys:
+                    # Use chunk-aligned reading for large data
+                    data[key] = self._read_key_chunk_aligned(zarr_group, key, futures)
+                else:
+                    # Use simple reading for other data
+                    data[key] = self._read_key_standard(zarr_group, key, futures)
+
+            # Wait for all futures to complete
+            for future in futures:
+                future.result()
+
+            return data
+
+    @profile
+    def _move_to_gpu(self, data: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]:
+        """Convert numpy arrays to torch tensors and move to GPU if available."""
+        result = {}
+
+        for key, array in data.items():
+            # Move to GPU if available
+            if self.dm.cuda:
+                result[key] = data[key].to(self.dm.device, non_blocking=True)
+
+        return result
+
+    def _convert_to_shard_tensors(
+        self, tensors: Dict[str, torch.Tensor]
+    ) -> Dict[str, Union[torch.Tensor, ShardTensor]]:
+        """Convert tensors to ShardTensor objects for distributed training."""
+        if self.device_mesh is None:
+            return tensors
+
+        result = {}
+
+        for key, tensor in tensors.items():
+
+            # Create a ShardTensor with whatever layout the data is actually in:
+            st = ShardTensor.__new__(
+                ShardTensor,
+                local_tensor=tensor,
+                spec=self.tensor_specs[key],
+                requires_grad=False,  # By default, the data pipe output doesn't need a grad.
+            )
+
+            # Find out the desired placement:
+            if isinstance(self.placements, dict):
+                target_placement = self.placements[key]
+            else:
+                target_placement = self.placements
+
+            # Redistribute if necessary:
+            # (Recall that this is one dimensional mesh only)
+            if st._spec.placements[0] != target_placement[0]:
+                st = st.redistribute(placements=target_placement)
+
+            result[key] = st
+
+        return result
+
+    @profile
+    def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
+        """
+        Get a data sample.
+
+        Args:
+            idx: Index of the sample to retrieve
+
+        Returns:
+            Dictionary containing tensors/ShardTensors for the requested data
+        """
+        if idx >= len(self.filenames):
+            raise IndexError(
+                f"Index {idx} out of range for dataset of size {len(self.filenames)}"
+            )
+
+        filename = self.filenames[idx]
+        filepath = self.data_path / filename
+
+        # Read data from zarr file
+        data = self._read_zarr_file(filepath)
+
+        # Convert to torch tensors
+        tensors = self._move_to_gpu(data)
+
+        # Convert to ShardTensors if using domain parallelism
+        if self.device_mesh is not None:
+            tensors = self._convert_to_shard_tensors(tensors)
+
+        return tensors
+
+
+# TODO: Additional features to consider implementing:
+# 1. Caching of metadata to avoid repeated zarr.open_group calls
+# 2. Asynchronous prefetching of next sample while processing current
+# 3. More sophisticated sharding strategies (e.g., spatial partitioning)
+# 4. Support for different placement strategies per key
+# 5. Memory-mapped reading for very large datasets
+# 6. Compression/decompression handling
+
+
+def create_transolver_dataset(
+    data_path: Union[str, Path],
+    device_mesh: Optional[DeviceMesh] = None,
+    placements: Optional[Tuple[Placement, ...]] = None,
+    **kwargs,
+) -> DomainParallelZarrDataset:
+    """
+    Factory function to create a Transolver dataset with sensible defaults.
+
+    Args:
+        data_path: Path to zarr data directory
+        device_mesh: Optional device mesh for domain parallelism
+        placements: Optional placement specifications
+        **kwargs: Additional arguments passed to DomainParallelZarrDataset
+
+    Returns:
+        Configured dataset instance
+    """
+    return DomainParallelZarrDataset(
+        data_path=data_path, device_mesh=device_mesh, placements=placements, **kwargs
+    )

From a07016572f288b9a844277cd2c03f86942b58d8d Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Wed, 2 Jul 2025 16:01:05 +0000
Subject: [PATCH 05/23] Updates for the transolver example using the
 consolidated and updated model code.

---
 examples/cfd/darcy_transolver/config_fix.yaml | 22 +++++++++----------
 .../train_transolver_darcy.py                 | 14 +++++++-----
 .../train_transolver_darcy_fix.py             | 19 ++++++++--------
 3 files changed, 28 insertions(+), 27 deletions(-)

diff --git a/examples/cfd/darcy_transolver/config_fix.yaml b/examples/cfd/darcy_transolver/config_fix.yaml
index ce24362bfb..e90fb50567 100644
--- a/examples/cfd/darcy_transolver/config_fix.yaml
+++ b/examples/cfd/darcy_transolver/config_fix.yaml
@@ -21,29 +21,29 @@
 # SOFTWARE.
 
 output_dir: ./output/darcy_transolver_fix
-run_id: fp32_dev_fullres_b64_s64-te
+run_id: fp16_dev_fullres_b8_s64-te
 
 data:
   train_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth1.npz
   test_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth2.npz
   resolution: 421 #421, 211, 141, 106, 85 all viable
-  batch_size: 32 # This is the GLOBAL batch size
+  batch_size: 8 # This is the GLOBAL batch size
 
 model:
-  space_dim: 2
+  functional_dim: 1
+  out_dim: 1
+  embedding_dim: 2
   n_layers: 4
-  n_hidden: 128
+  n_hidden: 64
   dropout: 0.0
-  n_head: 8
-  Time_Input: False
+  n_head: 4
   act: gelu
-  mlp_ratio: 1
-  fun_dim: 1
-  out_dim: 1
-  ref: 8
+  mlp_ratio: 2
   unified_pos: False
+  ref: 8
   slice_num: 64
-  transformer_engine: True
+  use_te: True
+  Time_Input: False
   
 precision: fp32
 
diff --git a/examples/cfd/darcy_transolver/train_transolver_darcy.py b/examples/cfd/darcy_transolver/train_transolver_darcy.py
index 325742abd0..734f5720b6 100644
--- a/examples/cfd/darcy_transolver/train_transolver_darcy.py
+++ b/examples/cfd/darcy_transolver/train_transolver_darcy.py
@@ -54,22 +54,24 @@ def darcy_trainer(cfg: DictConfig) -> None:
 
     # define model, loss, optimiser, scheduler, data loader
     model = Transolver(
-        space_dim=cfg.model.space_dim,
+        functional_dim=cfg.model.functional_dim,
+        out_dim=cfg.model.out_dim,
+        embedding_dim=cfg.model.embedding_dim,
         n_layers=cfg.model.n_layers,
         n_hidden=cfg.model.n_hidden,
         dropout=cfg.model.dropout,
         n_head=cfg.model.n_head,
-        Time_Input=cfg.model.Time_Input,
         act=cfg.model.act,
         mlp_ratio=cfg.model.mlp_ratio,
         fun_dim=cfg.model.fun_dim,
-        out_dim=cfg.model.out_dim,
         slice_num=cfg.model.slice_num,
-        ref=cfg.model.ref,
         unified_pos=cfg.model.unified_pos,
-        H=cfg.training.resolution,
-        W=cfg.training.resolution,
+        ref=cfg.model.ref,
+        structured_shape=[cfg.data.resolution, cfg.data.resolution],
+        use_te=cfg.model.use_te,
+        Time_Input=cfg.model.Time_Input,
     ).to(dist.device)
+
     loss_fun = TestLoss(size_average=False)
     optimizer = Adam(model.parameters(), lr=cfg.scheduler.initial_lr)
     scheduler = lr_scheduler.LambdaLR(
diff --git a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
index 2eea776208..2b6461b470 100644
--- a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
+++ b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
@@ -72,7 +72,7 @@ def forward_train_full_loop(
     """
     dm = DistributedManager()
     with context:
-        pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
+        pred = model(embedding=pos, fx=x.unsqueeze(-1)).squeeze(-1)
         pred = y_normalizer.decode(pred)
         loss = loss_fun(pred, y)
     if scaler is not None:
@@ -125,7 +125,7 @@ def val_epoch(model, test_dataloader, loss_fun, y_normalizer):
     for i, batch in enumerate(test_dataloader):
         pos, x, y = batch
         with torch.no_grad():
-            pred = model(pos, fx=x.unsqueeze(-1)).squeeze(-1)
+            pred = model(embedding=pos, fx=x.unsqueeze(-1)).squeeze(-1)
             pred = y_normalizer.decode(pred)
             loss = loss_fun(pred, y)
 
@@ -188,22 +188,21 @@ def darcy_trainer(cfg: DictConfig) -> None:
     # define model
     ########################################################################
     model = Transolver(
-        space_dim=cfg.model.space_dim,
+        functional_dim=cfg.model.functional_dim,
+        out_dim=cfg.model.out_dim,
+        embedding_dim=cfg.model.embedding_dim,
         n_layers=cfg.model.n_layers,
         n_hidden=cfg.model.n_hidden,
         dropout=cfg.model.dropout,
         n_head=cfg.model.n_head,
-        Time_Input=cfg.model.Time_Input,
         act=cfg.model.act,
         mlp_ratio=cfg.model.mlp_ratio,
-        fun_dim=cfg.model.fun_dim,
-        out_dim=cfg.model.out_dim,
         slice_num=cfg.model.slice_num,
-        ref=cfg.model.ref,
         unified_pos=cfg.model.unified_pos,
-        H=cfg.data.resolution,
-        W=cfg.data.resolution,
-        use_te=cfg.model.transformer_engine,
+        ref=cfg.model.ref,
+        structured_shape=[cfg.data.resolution, cfg.data.resolution],
+        use_te=cfg.model.use_te,
+        Time_Input=cfg.model.Time_Input,
     ).to(dm.device)
 
     if dm.world_size > 1:

From b1aa22fb722bf0e0b5dd9cfb8f87283636c54f84 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Wed, 2 Jul 2025 16:05:44 +0000
Subject: [PATCH 06/23] Add small function to convert matlab matrices to npz

---
 .../darcy_transolver/convert_mat_to_npz.py    | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)
 create mode 100644 examples/cfd/darcy_transolver/convert_mat_to_npz.py

diff --git a/examples/cfd/darcy_transolver/convert_mat_to_npz.py b/examples/cfd/darcy_transolver/convert_mat_to_npz.py
new file mode 100644
index 0000000000..44125a71f8
--- /dev/null
+++ b/examples/cfd/darcy_transolver/convert_mat_to_npz.py
@@ -0,0 +1,47 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: MIT License
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+import sys
+import numpy as np
+from scipy.io import loadmat
+
+
+def main(mat_file, npz_file):
+    # Load the .mat file
+    data = loadmat(mat_file)
+
+    # Extract 'coeff' and 'sol'
+    coeff = data["coeff"]
+    sol = data["sol"]
+
+    # Save to .npz file
+    np.savez(npz_file, coeff=coeff, sol=sol)
+    print(f"Saved 'coeff' and 'sol' to {npz_file}")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 2:
+        print("Usage: python convert_mat_to_npz.py input.mat")
+        sys.exit(1)
+    mat_file = sys.argv[1]
+    npz_file = mat_file.replace(".mat", ".npz")
+    main(mat_file, npz_file)

From 25a37ab8be2da2bf41b4682ddbfd4589f9974d06 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Wed, 2 Jul 2025 16:06:30 +0000
Subject: [PATCH 07/23] Updates to the transolver model architecture

---
 physicsnemo/models/transolver/transolver.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index eaa2c1f49a..6a50f13695 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -498,6 +498,7 @@ def __init__(
         self.structured_shape = structured_shape
 
         # If we're using the unified position, create and save the position embeddings:
+        self.unified_pos = unified_pos
 
         if unified_pos:
             if structured_shape is None:

From 4c383a3f6773c819d20ac92b3dd1df54e18f169f Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Tue, 8 Jul 2025 19:56:18 +0000
Subject: [PATCH 08/23] Update transolver model.  Add external aerodynamics
 example with transolver.

---
 .../transolver/conf/train.yaml                |  82 ++++
 .../external_aerodynamics/transolver/loss.py  |  58 +++
 .../transolver/metrics.py                     |  64 +++
 .../external_aerodynamics/transolver/train.py | 439 ++++++++++++++++++
 .../models/transolver/Physics_Attention.py    | 278 +----------
 physicsnemo/models/transolver/transolver.py   | 268 +----------
 6 files changed, 653 insertions(+), 536 deletions(-)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/conf/train.yaml
 create mode 100644 examples/cfd/external_aerodynamics/transolver/loss.py
 create mode 100644 examples/cfd/external_aerodynamics/transolver/metrics.py
 create mode 100644 examples/cfd/external_aerodynamics/transolver/train.py

diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train.yaml
new file mode 100644
index 0000000000..37dc470d9e
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train.yaml
@@ -0,0 +1,82 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+
+output_dir: "outputs"
+run_id: "transolver_te-1cycle"
+
+# Training configuration
+training:
+  num_epochs: 400
+  save_interval: 25  # Save checkpoint every N epochs
+  seed: 42
+  output_dir: "outputs"
+
+# Model configuration
+model:
+  _target_: physicsnemo.models.transolver.Transolver
+  functional_dim: 2 # Input feature dimension
+  out_dim: 4        # Output feature dimension
+  embedding_dim: 6  # Spatial embedding dimension
+  n_layers: 8       # Number of transformer layers
+  n_hidden: 256     # Hidden dimension
+  dropout: 0.0      # Dropout rate
+  n_head: 8         # Number of attention heads
+  act: "gelu"       # Activation function
+  mlp_ratio: 4      # MLP ratio in attention blocks
+  slice_num: 128     # Number of slices in physics attention
+  unified_pos: false # Whether to use unified positional embeddings
+  ref: 8            # Reference dimension for unified pos
+  structured_shape: null
+  use_te: true      # Use transformer engine
+  Time_Input: false # Whether to use time embeddings
+
+
+
+# Optimizer configuration
+optimizer:
+  _target_: torch.optim.AdamW
+  lr: 1.0e-3
+  weight_decay: 1.0e-4
+  betas: [0.9, 0.999]
+  eps: 1.0e-8
+
+# Data configuration
+data:
+  train:
+    data_path: /group_data/datasets/drivaer_aws/domino/train/
+  val:
+    data_path: /group_data/datasets/drivaer_aws/domino/val/
+  max_workers: 32
+  pin_memory: true
+  resolution: 262_144
+  surface_keys:
+    - "surface_fields"
+    - "surface_mesh_centers"
+    - "surface_normals"
+    - "air_density"
+    - "stream_velocity"
+
+# Logging configuration
+logging:
+  level: INFO
+  format: '%(asctime)s - %(name)s - %(levelname)s - %(message)s'
+
+# Hardware configuration
+hardware:
+  precision: float32
+  compile_model: false  # Whether to use torch.compile
\ No newline at end of file
diff --git a/examples/cfd/external_aerodynamics/transolver/loss.py b/examples/cfd/external_aerodynamics/transolver/loss.py
new file mode 100644
index 0000000000..3420d7324c
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/loss.py
@@ -0,0 +1,58 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import torch
+from typing import Literal
+
+
+def loss_fn(pred, target):
+    return loss_fn_surface(pred, target, "mse")
+
+
+def loss_fn_surface(
+    output: torch.Tensor, target: torch.Tensor, loss_type: Literal["mse", "rmse"]
+) -> torch.Tensor:
+    """Calculate loss for surface data by handling scalar and vector components separately.
+
+    Args:
+        output: Predicted surface values from the model
+        target: Ground truth surface values
+        loss_type: Type of loss to calculate ("mse" or "rmse")
+
+    Returns:
+        Combined scalar and vector loss as a scalar tensor
+    """
+    # Separate the scalar and vector components:
+    output_pressure, output_sheer = torch.split(output, [1, 3], dim=2)
+    target_pressure, target_sheer = torch.split(target, [1, 3], dim=2)
+
+    numerator_pressure = torch.mean((output_pressure - target_pressure) ** 2.0)
+    numerator_sheer = torch.mean((target_sheer - output_sheer) ** 2.0, (0, 1))
+
+    if loss_type == "mse":
+        loss_pressure = numerator_pressure
+        loss_wall_sheer = torch.sum(numerator_sheer)
+    else:
+        denom = torch.mean((target_pressure) ** 2.0)
+        loss_pressure = numerator_pressure / denom
+
+        # Compute the mean diff**2 of the vector component, leave the last dimension:
+        denom_sheer = torch.mean((target_sheer) ** 2.0, (0, 1))
+        loss_wall_sheer = torch.sum(numerator_sheer / denom_sheer)
+
+    loss = loss_pressure + loss_wall_sheer
+
+    return loss / 4.0
diff --git a/examples/cfd/external_aerodynamics/transolver/metrics.py b/examples/cfd/external_aerodynamics/transolver/metrics.py
new file mode 100644
index 0000000000..5bba878e2a
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/metrics.py
@@ -0,0 +1,64 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import torch
+import torch.distributed as dist
+
+from physicsnemo.distributed import DistributedManager
+
+
+def all_reduce_dict(metrics, dm):
+    # TODO - update this to use domains and not the full world
+
+    if dm.world_size == 1:
+        return metrics
+
+    for key, value in metrics.items():
+        dist.all_reduce(value)
+        value = value / dm.world_size
+        metrics[key] = value
+
+    return metrics
+
+
+def metrics_fn(pred, target, dm):
+    return metrics_fn_surface(pred, target, dm)
+
+
+def metrics_fn_surface(pred, target, dm):
+
+    l2_num = (pred - target) ** 2
+    l2_num = torch.sum(l2_num, dim=1)
+    l2_num = torch.sqrt(l2_num)
+
+    l2_denom = target**2
+    l2_denom = torch.sum(l2_denom, dim=1)
+    l2_denom = torch.sqrt(l2_denom)
+
+    l2 = l2_num / l2_denom
+
+    metrics = {
+        "l2_pressure": torch.mean(l2[:, 0]),
+        "l2_sheer_x": torch.mean(l2[:, 1]),
+        "l2_sheer_y": torch.mean(l2[:, 2]),
+        "l2_sheer_z": torch.mean(l2[:, 3]),
+    }
+
+    return metrics
+
+
+def metrics_fn_surface_pressure(pred, target):
+    return torch.mean((pred - target) ** 2.0)
diff --git a/examples/cfd/external_aerodynamics/transolver/train.py b/examples/cfd/external_aerodynamics/transolver/train.py
new file mode 100644
index 0000000000..12b2e78708
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/train.py
@@ -0,0 +1,439 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import time
+import torch
+import hydra
+import omegaconf
+from tabulate import tabulate
+from omegaconf import DictConfig
+from torch.utils.tensorboard import SummaryWriter
+
+from physicsnemo.launch.utils import load_checkpoint, save_checkpoint
+from physicsnemo.launch.logging import PythonLogger, RankZeroLoggingWrapper
+from physicsnemo.distributed import DistributedManager
+from typing import Literal
+
+from physicsnemo.utils.profiling import profile, Profiler
+
+from datapipe import DomainParallelZarrDataset
+from loss import loss_fn
+from metrics import metrics_fn
+
+
+@profile
+def preprocess_data(batch):
+
+    mesh_centers = batch["surface_mesh_centers"]
+    normals = batch["surface_normals"]
+    targets = batch["surface_fields"]
+    node_features = torch.stack(
+        [batch["air_density"], batch["stream_velocity"]], dim=-1
+    ).to(torch.float32)
+
+    # fourier_sin_features = [
+    #     torch.sin(mesh_centers * (2 ** i) * torch.pi)
+    #     for i in range(4)
+    # ]
+    # fourier_cos_features = [
+    #     torch.cos(mesh_centers * (2 ** i) * torch.pi)
+    #     for i in range(4)
+    # ]
+
+    embeddings = torch.cat(
+        [
+            mesh_centers,
+            normals,
+            # *fourier_sin_features,
+            # *fourier_cos_features
+        ],
+        dim=-1,
+    )
+
+    node_features = node_features.unsqueeze(0).broadcast_to(embeddings.shape[0], -1)
+
+    return node_features, embeddings, targets
+
+
+@profile
+def downsample(features, embeddings, targets, num_keep=1024):
+    # Determine the number of samples to keep (e.g., 50% of original size)
+    num_samples = features.shape[0]
+
+    # Generate random indices to keep
+    indices = torch.randperm(num_samples)[:num_keep]
+
+    # Use the same indices to downsample all tensors
+    downsampled_features = features[indices]
+    downsampled_embeddings = embeddings[indices]
+    downsampled_targets = targets[indices]
+
+    downsampled_features = downsampled_features.unsqueeze(0)
+    downsampled_embeddings = downsampled_embeddings.unsqueeze(0)
+    downsampled_targets = downsampled_targets.unsqueeze(0)
+
+    return downsampled_features, downsampled_embeddings, downsampled_targets
+
+
+@profile
+def train_epoch(
+    dataloader,
+    sampler,
+    model,
+    optimizer,
+    scheduler,
+    logger,
+    writer,
+    epoch,
+    cfg,
+    dist_manager,
+):
+    """Train for one epoch
+
+    Args:
+        dataloader: Training data loader
+        model: The model to train
+        logger: Python logger instance
+        writer: Tensorboard writer
+        cfg: Configuration object
+    """
+    model.train()
+    total_loss = 0
+    total_metrics = {}
+
+    epoch_indices = list(sampler)
+    epoch_len = len(epoch_indices)
+
+    start_time = time.time()
+    for i, batch_idx in enumerate(epoch_indices):
+        batch = dataloader[batch_idx]
+        # Get data from batch
+        features, embeddings, targets = preprocess_data(batch)
+
+        features, embeddings, targets = downsample(
+            features, embeddings, targets, cfg.data.resolution
+        )
+
+        # Forward pass
+        outputs = model(features, embeddings)
+
+        metrics = metrics_fn(outputs, targets, dist_manager)
+
+        if i == 0:
+            total_metrics = metrics
+        else:
+            total_metrics = {
+                k: total_metrics[k] + metrics[k].item() for k in metrics.keys()
+            }
+
+        loss = loss_fn(outputs, targets)
+
+        # Backward pass
+        optimizer.zero_grad()
+        loss.backward()
+        optimizer.step()
+        scheduler.step()
+
+        end_time = time.time()
+        duration = end_time - start_time
+        start_time = end_time
+
+        images_per_second = 1 / duration
+
+        # Logging
+        total_loss += loss.item()
+
+        logger.info(
+            f"Epoch {epoch} [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
+        )
+        if dist_manager.rank == 0:
+            writer.add_scalar(
+                "batch/learning_rate",
+                optimizer.param_groups[0]["lr"],
+                i + epoch_len * epoch,
+            )
+            writer.add_scalar("batch/loss", loss.item(), i + epoch_len * epoch)
+            writer.add_scalar(
+                "batch/throughpu_per_gpu", images_per_second, i + epoch_len * epoch
+            )
+            for metric_name, metric_value in metrics.items():
+                writer.add_scalar(
+                    f"batch/{metric_name}", metric_value, i + epoch_len * epoch
+                )
+
+    avg_loss = total_loss / epoch_len
+    avg_metrics = {k: v / epoch_len for k, v in total_metrics.items()}
+    if dist_manager.rank == 0:
+        writer.add_scalar("epoch/loss", avg_loss, epoch)
+        for metric_name, metric_value in avg_metrics.items():
+            writer.add_scalar(f"epoch/{metric_name}", metric_value, epoch)
+        # Print average metrics using tabulate
+        metrics_table = tabulate(
+            [[k, v] for k, v in avg_metrics.items()],
+            headers=["Metric", "Average Value"],
+            tablefmt="pretty",
+        )
+        print(f"\nEpoch {epoch} Average Metrics:\n{metrics_table}\n")
+    return avg_loss
+
+
+@profile
+def val_epoch(dataloader, sampler, model, logger, val_writer, epoch, cfg, dist_manager):
+    """Validation for one epoch
+
+    Args:
+        dataloader: Validation data loader
+        sampler: Validation data sampler
+        model: The model to evaluate
+        logger: Python logger instance
+        writer: Tensorboard writer
+        epoch: Current epoch number
+        cfg: Configuration object
+        dist_manager: Distributed manager instance
+    """
+
+    model.eval()  # Set model to evaluation mode
+    total_loss = 0
+    total_metrics = {}
+
+    epoch_indices = list(sampler)
+    epoch_len = len(epoch_indices)
+
+    start_time = time.time()
+    with torch.no_grad():  # Disable gradient computation
+        for i, batch_idx in enumerate(epoch_indices):
+            batch = dataloader[batch_idx]
+            # Get data from batch
+            features, embeddings, targets = preprocess_data(batch)
+
+            features, embeddings, targets = downsample(
+                features, embeddings, targets, cfg.data.resolution
+            )
+
+            # Forward pass
+            outputs = model(features, embeddings)
+
+            metrics = metrics_fn(outputs, targets, dist_manager)
+
+            if i == 0:
+                total_metrics = metrics
+            else:
+                total_metrics = {
+                    k: total_metrics[k] + metrics[k].item() for k in metrics.keys()
+                }
+
+            loss = loss_fn(outputs, targets)
+
+            end_time = time.time()
+            duration = end_time - start_time
+            start_time = end_time
+
+            # Logging
+            total_loss += loss.item()
+
+            logger.info(
+                f"Val [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
+            )
+            # We don't add individual loss measurements in the validation loop.
+
+    avg_loss = total_loss / epoch_len
+    avg_metrics = {k: v / epoch_len for k, v in total_metrics.items()}
+    if dist_manager.rank == 0:
+        val_writer.add_scalar("epoch/loss", avg_loss, epoch)
+        for metric_name, metric_value in avg_metrics.items():
+            val_writer.add_scalar(f"epoch/{metric_name}", metric_value, epoch)
+        # Print average metrics using tabulate
+        metrics_table = tabulate(
+            [[k, v] for k, v in avg_metrics.items()],
+            headers=["Metric", "Average Value"],
+            tablefmt="pretty",
+        )
+        print(f"\nEpoch {epoch} Validation Average Metrics:\n{metrics_table}\n")
+    return avg_loss
+
+
+@profile
+def main(cfg: DictConfig):
+    """Main training function
+
+    Args:
+        cfg: Hydra configuration object
+    """
+
+    DistributedManager.initialize()
+
+    # Set up distributed training
+    dist_manager = DistributedManager()
+
+    # Set up logging
+    logger = RankZeroLoggingWrapper(PythonLogger(name="training"), dist_manager)
+    if dist_manager.rank == 0:
+        os.makedirs(cfg.output_dir, exist_ok=True)
+        writer = SummaryWriter(
+            log_dir=os.path.join(
+                cfg.output_dir + "/" + cfg.run_id + "/train",
+            )
+        )
+        val_writer = SummaryWriter(
+            log_dir=os.path.join(
+                cfg.output_dir + "/" + cfg.run_id + "/val",
+            )
+        )
+    else:
+        writer = None
+        val_writer = None
+
+    logger.info(f"Config:\n{omegaconf.OmegaConf.to_yaml(cfg, resolve=True)}")
+
+    # Set up model
+    model = hydra.utils.instantiate(cfg.model)
+    model.to(dist_manager.device)
+
+    if dist_manager.world_size > 1:
+        model = torch.nn.parallel.DistributedDataParallel(
+            model,
+            device_ids=[dist_manager.local_rank],
+            output_device=dist_manager.device,
+        )
+
+    num_params = sum(p.numel() for p in model.parameters())
+    logger.info(f"Number of parameters: {num_params}")
+
+    # Set up data
+    device_mesh = None
+    placements = None
+
+    # Training dataset
+    train_dataset = DomainParallelZarrDataset(
+        data_path=cfg.data.train.data_path,
+        device_mesh=device_mesh,
+        placements=placements,
+        max_workers=cfg.data.max_workers,
+        pin_memory=cfg.data.pin_memory,
+        keys_to_read=cfg.data.surface_keys,
+        large_keys=None,
+    )
+
+    # Validation dataset
+    val_dataset = DomainParallelZarrDataset(
+        data_path=cfg.data.val.data_path,  # Assuming validation data path is configured
+        device_mesh=device_mesh,
+        placements=placements,
+        max_workers=cfg.data.max_workers,
+        pin_memory=cfg.data.pin_memory,
+        keys_to_read=cfg.data.surface_keys,
+        large_keys=None,
+    )
+
+    # Set up distributed samplers
+    train_sampler = torch.utils.data.distributed.DistributedSampler(
+        train_dataset,
+        num_replicas=dist_manager.world_size,
+        rank=dist_manager.rank,
+        shuffle=True,
+        drop_last=True,
+    )
+
+    val_sampler = torch.utils.data.distributed.DistributedSampler(
+        val_dataset,
+        num_replicas=dist_manager.world_size,
+        rank=dist_manager.rank,
+        shuffle=False,  # No shuffling for validation
+        drop_last=True,
+    )
+
+    # Set up optimizer and scheduler
+    optimizer = hydra.utils.instantiate(cfg.optimizer, params=model.parameters())
+    # Set up OneCycleLR learning rate scheduler
+    scheduler = torch.optim.lr_scheduler.OneCycleLR(
+        optimizer,
+        max_lr=cfg.optimizer.lr,
+        total_steps=len(list(train_sampler)) * cfg.training.num_epochs,
+        pct_start=0.3,
+        div_factor=25.0,
+        final_div_factor=1e4,
+    )
+
+    ckpt_args = {
+        "path": f"{cfg.output_dir}/runs/{cfg.run_id}/checkpoints",
+        "optimizer": optimizer,
+        "scheduler": scheduler,
+        "models": model,
+    }
+    loaded_epoch = load_checkpoint(device=dist_manager.device, **ckpt_args)
+
+    # Training loop
+    logger.info("Starting training...")
+    for epoch in range(loaded_epoch, cfg.training.num_epochs):
+        # Set the epoch in the samplers
+        train_sampler.set_epoch(epoch)
+        val_sampler.set_epoch(epoch)
+
+        # Training phase
+        train_loss = train_epoch(
+            train_dataset,
+            train_sampler,
+            model,
+            optimizer,
+            scheduler,
+            logger,
+            writer,
+            epoch,
+            cfg,
+            dist_manager,
+        )
+
+        # Validation phase
+        val_loss = val_epoch(
+            val_dataset,
+            val_sampler,
+            model,
+            logger,
+            val_writer,
+            epoch,
+            cfg,
+            dist_manager,
+        )
+
+        # Log epoch results
+        logger.info(
+            f"Epoch [{epoch}/{cfg.training.num_epochs}] Train Loss: {train_loss:.6f} Val Loss: {val_loss:.6f}"
+        )
+
+        # save checkpoint
+        if epoch % cfg.training.save_interval == 0 and dist_manager.rank == 0:
+            save_checkpoint(**ckpt_args, epoch=epoch)
+
+    logger.info("Training completed!")
+
+
+@hydra.main(version_base=None, config_path="conf", config_name="train")
+def launch(cfg: DictConfig):
+    """Launch training with hydra configuration
+
+    Args:
+        cfg: Hydra configuration object
+    """
+    # profiler = Profiler()
+    # profiler.enable("line_profiler")
+    # profiler.initialize()
+    main(cfg)
+    # profiler.finalize()
+
+
+if __name__ == "__main__":
+    launch()
diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index c303634391..ab1409eb21 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -67,7 +67,6 @@ def __init__(self, dim, heads, dim_head, dropout, slice_num, use_te):
         self.in_project_slice = nn.Linear(dim_head, slice_num)
         for l_i in [self.in_project_slice]:
             torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
-
         if not use_te:
             self.to_q = nn.Linear(dim_head, dim_head, bias=False)
             self.to_k = nn.Linear(dim_head, dim_head, bias=False)
@@ -228,61 +227,31 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
 
         # Project the inputs onto learned spaces:
         x_mid, fx_mid = self.project_input_onto_slices(x)
-        # if (x_mid.isnan().any()):
-        #     print(f"Rank {dm.rank} x_mid is nan")
-        #     exit()
-        # if (fx_mid.isnan().any()):
-        #     print(f"Rank {dm.rank} fx_mid is nan")
-        #     exit()
-        # x_mid and fx_mid should have shapes of [B, N_head, N_tokens, Head_dim]
 
         # Perform the linear projection of learned latent space onto slices:
         slice_projections = self.in_project_slice(x_mid)
-        # if (slice_projections.isnan().any()):
-        #     print(f"Rank {dm.rank} slice_projections is nan")
-        #     exit()
+
         # Slice projections has shape [B, N_head, N_tokens, Head_dim], but head_dim may have changed!
 
         # Use the slice projections and learned spaces to compute the slices, and their weights:
         slice_weights, slice_tokens = self.compute_slices_from_projections(
             slice_projections, fx_mid
         )
-        # if (slice_weights.isnan().any()):
-        #     print(f"Rank {dm.rank} slice_weights is nan")
-        #     exit()
-        # if (slice_tokens.isnan().any()) or (slice_tokens.isinf().any()):
-        #     print(f"Rank {dm.rank} slice_tokens is nan or inf")
-        #     torch.save(slice_tokens, f"slice_tokens_{dm.rank}.pt")
-        #     torch.save(slice_weights, f"slice_weights_{dm.rank}.pt")
-        #     torch.save(slice_projections, f"slice_projections_{dm.rank}.pt")
-        #     torch.save(fx_mid, f"fx_mid_{dm.rank}.pt")
-        #     exit()
         # slice_weights has shape [Batch, N_heads, N_tokens, Slice_num]
         # slice_tokens has shape  [Batch, N_heads, N_tokens, head_dim]
 
-        # print(f"Rank {dm.rank} Min/max/std of slice tokens: {slice_tokens.min()}, {slice_tokens.max()}, {slice_tokens.std()}")
-
         # Apply attention to the slice tokens
         if self.use_te:
             out_slice_token = self.compute_slice_attention_te(slice_tokens)
-            # out_slice_token = self.compute_slice_attention(slice_tokens)
         else:
+            # out_slice_token = self.compute_slice_attention(slice_tokens)
             out_slice_token = self.compute_slice_attention_sdpa(slice_tokens)
 
-        # if (out_slice_token.isnan().any()):
-        #     print(f"Rank {dm.rank} out_slice_token is nan")
-        #     torch.save(slice_tokens, f"slice_tokens_{dm.rank}.pt")
-        #     torch.save(slice_weights, f"slice_weights_{dm.rank}.pt")
-        #     torch.save(slice_projections, f"slice_projections_{dm.rank}.pt")
-        #     torch.save(fx_mid, f"fx_mid_{dm.rank}.pt")
-        #     exit()
         # Shape unchanged
 
         # Deslice:
         outputs = self.project_attention_outputs(out_slice_token, slice_weights)
-        # if (outputs.isnan().any()):
-        #     print(f"Rank {dm.rank} outputs is nan")
-        #     exit()
+
         # Outputs now has the same shape as the original input x
 
         return outputs
@@ -298,7 +267,6 @@ def __init__(
     ):
         super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
         inner_dim = dim_head * heads
-
         self.in_project_x = nn.Linear(dim, inner_dim)
         self.in_project_fx = nn.Linear(dim, inner_dim)
 
@@ -306,7 +274,6 @@ def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
         """
         Project the input onto the slice space.
         """
-
         fx_mid = rearrange(
             self.in_project_fx(x), "B N (h d) -> B h N d", h=self.heads, d=self.dim_head
         )
@@ -421,242 +388,3 @@ def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
         )
 
         return input_projected_x, input_projected_fx
-
-
-class Physics_Attention_Irregular_Mesh(nn.Module):
-    "for irregular meshes in 1D, 2D or 3D space"
-
-    def __init__(self, dim, heads=8, dim_head=64, dropout=0.0, slice_num=64):
-        super().__init__()
-        inner_dim = dim_head * heads
-        self.dim_head = dim_head
-        self.heads = heads
-        self.scale = dim_head**-0.5
-        self.softmax = nn.Softmax(dim=-1)
-        self.dropout = nn.Dropout(dropout)
-        self.temperature = nn.Parameter(torch.ones([1, heads, 1, 1]) * 0.5)
-
-        self.in_project_x = nn.Linear(dim, inner_dim)
-        self.in_project_fx = nn.Linear(dim, inner_dim)
-        self.in_project_slice = nn.Linear(dim_head, slice_num)
-        for l_i in [self.in_project_slice]:
-            torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
-        self.to_q = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_k = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_v = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
-
-    def forward(self, x):
-        # B N C
-        B, N, C = x.shape
-
-        ### (1) Slice
-        fx_mid = (
-            self.in_project_fx(x)
-            .reshape(B, N, self.heads, self.dim_head)
-            .permute(0, 2, 1, 3)
-            .contiguous()
-        )  # B H N C
-        x_mid = (
-            self.in_project_x(x)
-            .reshape(B, N, self.heads, self.dim_head)
-            .permute(0, 2, 1, 3)
-            .contiguous()
-        )  # B H N C
-        slice_weights = self.softmax(
-            self.in_project_slice(x_mid) / self.temperature
-        )  # B H N G
-        slice_norm = slice_weights.sum(2)  # B H G
-        slice_token = torch.einsum("bhnc,bhng->bhgc", fx_mid, slice_weights)
-        slice_token = slice_token / (
-            (slice_norm + 1e-5)[:, :, :, None].repeat(1, 1, 1, self.dim_head)
-        )
-
-        ### (2) Attention among slice tokens
-        q_slice_token = self.to_q(slice_token)
-        k_slice_token = self.to_k(slice_token)
-        v_slice_token = self.to_v(slice_token)
-        dots = torch.matmul(q_slice_token, k_slice_token.transpose(-1, -2)) * self.scale
-        attn = self.softmax(dots)
-        attn = self.dropout(attn)
-        out_slice_token = torch.matmul(attn, v_slice_token)  # B H G D
-
-        ### (3) Deslice
-        out_x = torch.einsum("bhgc,bhng->bhnc", out_slice_token, slice_weights)
-        out_x = rearrange(out_x, "b h n d -> b n (h d)")
-        return self.to_out(out_x)
-
-
-class Physics_Attention_Structured_Mesh_2D(nn.Module):
-    "for structured mesh in 2D space"
-
-    def __init__(
-        self,
-        dim,
-        heads=8,
-        dim_head=64,
-        dropout=0.0,
-        slice_num=64,
-        H=101,
-        W=31,
-        kernel=3,
-    ):  # kernel=3):
-        super().__init__()
-        inner_dim = dim_head * heads
-        self.dim_head = dim_head
-        self.heads = heads
-        self.scale = dim_head**-0.5
-        self.softmax = nn.Softmax(dim=-1)
-        self.dropout = nn.Dropout(dropout)
-        self.temperature = nn.Parameter(torch.ones([1, heads, 1, 1]) * 0.5)
-        self.H = H
-        self.W = W
-
-        self.in_project_x = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
-        self.in_project_fx = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
-        self.in_project_slice = nn.Linear(dim_head, slice_num)
-        for l_i in [self.in_project_slice]:
-            torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
-        self.to_q = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_k = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_v = nn.Linear(dim_head, dim_head, bias=False)
-
-        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
-
-    def forward(self, x):
-        # B N C
-        B, N, C = x.shape
-        x = (
-            x.reshape(B, self.H, self.W, C)
-            .contiguous()
-            .permute(0, 3, 1, 2)
-            .contiguous()
-        )  # B C H W
-
-        ### (1) Slice
-        fx_mid = (
-            self.in_project_fx(x)
-            .permute(0, 2, 3, 1)
-            .contiguous()
-            .reshape(B, N, self.heads, self.dim_head)
-            .permute(0, 2, 1, 3)
-            .contiguous()
-        )  # B H N C
-        x_mid = (
-            self.in_project_x(x)
-            .permute(0, 2, 3, 1)
-            .contiguous()
-            .reshape(B, N, self.heads, self.dim_head)
-            .permute(0, 2, 1, 3)
-            .contiguous()
-        )  # B H N G
-        slice_weights = self.softmax(
-            self.in_project_slice(x_mid) / torch.clamp(self.temperature, min=0.1, max=5)
-        )  # B H N G
-        slice_norm = slice_weights.sum(2)  # B H G
-        slice_token = torch.einsum("bhnc,bhng->bhgc", fx_mid, slice_weights)
-        slice_token = slice_token / (
-            (slice_norm + 1e-5)[:, :, :, None].repeat(1, 1, 1, self.dim_head)
-        )
-
-        ### (2) Attention among slice tokens
-        q_slice_token = self.to_q(slice_token)
-        k_slice_token = self.to_k(slice_token)
-        v_slice_token = self.to_v(slice_token)
-        dots = torch.matmul(q_slice_token, k_slice_token.transpose(-1, -2)) * self.scale
-        attn = self.softmax(dots)
-        attn = self.dropout(attn)
-        out_slice_token = torch.matmul(attn, v_slice_token)  # B H G D
-
-        ### (3) Deslice
-        out_x = torch.einsum("bhgc,bhng->bhnc", out_slice_token, slice_weights)
-        out_x = rearrange(out_x, "b h n d -> b n (h d)")
-        return self.to_out(out_x)
-
-
-class Physics_Attention_Structured_Mesh_3D(nn.Module):
-    "for structured mesh in 3D space"
-
-    def __init__(
-        self,
-        dim,
-        heads=8,
-        dim_head=64,
-        dropout=0.0,
-        slice_num=32,
-        H=32,
-        W=32,
-        D=32,
-        kernel=3,
-    ):
-        super().__init__()
-        inner_dim = dim_head * heads
-        self.dim_head = dim_head
-        self.heads = heads
-        self.scale = dim_head**-0.5
-        self.softmax = nn.Softmax(dim=-1)
-        self.dropout = nn.Dropout(dropout)
-        self.temperature = nn.Parameter(torch.ones([1, heads, 1, 1]) * 0.5)
-        self.H = H
-        self.W = W
-        self.D = D
-
-        self.in_project_x = nn.Conv3d(dim, inner_dim, kernel, 1, kernel // 2)
-        self.in_project_fx = nn.Conv3d(dim, inner_dim, kernel, 1, kernel // 2)
-        self.in_project_slice = nn.Linear(dim_head, slice_num)
-        for l_i in [self.in_project_slice]:
-            torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
-        self.to_q = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_k = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_v = nn.Linear(dim_head, dim_head, bias=False)
-        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
-
-    def forward(self, x):
-        # B N C
-        B, N, C = x.shape
-        x = (
-            x.reshape(B, self.H, self.W, self.D, C)
-            .contiguous()
-            .permute(0, 4, 1, 2, 3)
-            .contiguous()
-        )  # B C H W
-
-        ### (1) Slice
-        fx_mid = (
-            self.in_project_fx(x)
-            .permute(0, 2, 3, 4, 1)
-            .contiguous()
-            .reshape(B, N, self.heads, self.dim_head)
-            .permute(0, 2, 1, 3)
-            .contiguous()
-        )  # B H N C
-        x_mid = (
-            self.in_project_x(x)
-            .permute(0, 2, 3, 4, 1)
-            .contiguous()
-            .reshape(B, N, self.heads, self.dim_head)
-            .permute(0, 2, 1, 3)
-            .contiguous()
-        )  # B H N G
-        slice_weights = self.softmax(
-            self.in_project_slice(x_mid) / torch.clamp(self.temperature, min=0.1, max=5)
-        )  # B H N G
-        slice_norm = slice_weights.sum(2)  # B H G
-        slice_token = torch.einsum("bhnc,bhng->bhgc", fx_mid, slice_weights)
-        slice_token = slice_token / (
-            (slice_norm + 1e-5)[:, :, :, None].repeat(1, 1, 1, self.dim_head)
-        )
-
-        ### (2) Attention among slice tokens
-        q_slice_token = self.to_q(slice_token)
-        k_slice_token = self.to_k(slice_token)
-        v_slice_token = self.to_v(slice_token)
-        dots = torch.matmul(q_slice_token, k_slice_token.transpose(-1, -2)) * self.scale
-        attn = self.softmax(dots)
-        attn = self.dropout(attn)
-        out_slice_token = torch.matmul(attn, v_slice_token)  # B H G D
-
-        ### (3) Deslice
-        out_x = torch.einsum("bhgc,bhng->bhnc", out_slice_token, slice_weights)
-        out_x = rearrange(out_x, "b h n d -> b n (h d)")
-        return self.to_out(out_x)
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index 6a50f13695..6dbf16d3f7 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -205,153 +205,6 @@ def forward(self, fx):
             return fx
 
 
-# class Model(nn.Module):
-#     def __init__(
-#         self,
-#         space_dim=1,
-#         n_layers=5,
-#         n_hidden=256,
-#         dropout=0.0,
-#         n_head=8,
-#         Time_Input=False,
-#         act="gelu",
-#         mlp_ratio=1,
-#         fun_dim=1,
-#         out_dim=1,
-#         slice_num=32,
-#         ref=8,
-#         unified_pos=False,
-#         H=85,
-#         W=85,
-#         use_te=True,
-#     ):
-#         super().__init__()
-#         self.__name__ = "Transolver_2D"
-#         self.H = H
-#         self.W = W
-#         self.ref = ref
-#         self.unified_pos = unified_pos
-#         if self.unified_pos:
-#             self.pos = self.get_grid()
-#             self.preprocess = MLP(
-#                 fun_dim + self.ref * self.ref,
-#                 n_hidden * 2,
-#                 n_hidden,
-#                 n_layers=0,
-#                 res=False,
-#                 act=act,
-#             )
-#         else:
-#             self.preprocess = MLP(
-#                 fun_dim + space_dim,
-#                 n_hidden * 2,
-#                 n_hidden,
-#                 n_layers=0,
-#                 res=False,
-#                 act=act,
-#             )
-
-#         self.Time_Input = Time_Input
-#         self.n_hidden = n_hidden
-#         self.space_dim = space_dim
-#         if Time_Input:
-#             self.time_fc = nn.Sequential(
-#                 nn.Linear(n_hidden, n_hidden), nn.SiLU(), nn.Linear(n_hidden, n_hidden)
-#             )
-#         self.blocks = nn.ModuleList(
-#             [
-#                 Transolver_block(
-#                     num_heads=n_head,
-#                     hidden_dim=n_hidden,
-#                     dropout=dropout,
-#                     act=act,
-#                     mlp_ratio=mlp_ratio,
-#                     out_dim=out_dim,
-#                     slice_num=slice_num,
-#                     H=H,
-#                     W=W,
-#                     last_layer=(_ == n_layers - 1),
-#                     use_te=use_te,
-#                 )
-#                 for _ in range(n_layers)
-#             ]
-#         )
-#         self.initialize_weights()
-#         # self.placeholder = nn.Parameter(
-#         #     (1 / (n_hidden)) * torch.rand(n_hidden, dtype=torch.float)
-#         # )
-
-#     def initialize_weights(self):
-#         self.apply(self._init_weights)
-
-#     def _init_weights(self, m):
-#         if isinstance(m, nn.Linear):
-#             nn.init.trunc_normal_(m.weight, std=0.02)
-#             if isinstance(m, nn.Linear) and m.bias is not None:
-#                 nn.init.constant_(m.bias, 0)
-#         elif isinstance(m, (nn.LayerNorm, te.LayerNorm, nn.BatchNorm1d)):
-#             nn.init.constant_(m.bias, 0)
-#             nn.init.constant_(m.weight, 1.0)
-
-#     def get_grid(self, batchsize=1):
-#         size_x, size_y = self.H, self.W
-#         gridx = torch.tensor(np.linspace(0, 1, size_x), dtype=torch.float)
-#         gridx = gridx.reshape(1, size_x, 1, 1).repeat([batchsize, 1, size_y, 1])
-#         gridy = torch.tensor(np.linspace(0, 1, size_y), dtype=torch.float)
-#         gridy = gridy.reshape(1, 1, size_y, 1).repeat([batchsize, size_x, 1, 1])
-#         grid = torch.cat((gridx, gridy), dim=-1)  # B H W 2
-
-#         gridx = torch.tensor(np.linspace(0, 1, self.ref), dtype=torch.float)
-#         gridx = gridx.reshape(1, self.ref, 1, 1).repeat([batchsize, 1, self.ref, 1])
-#         gridy = torch.tensor(np.linspace(0, 1, self.ref), dtype=torch.float)
-#         gridy = gridy.reshape(1, 1, self.ref, 1).repeat([batchsize, self.ref, 1, 1])
-#         grid_ref = torch.cat((gridx, gridy), dim=-1)  # B H W 8 8 2
-
-#         pos = (
-#             torch.sqrt(
-#                 torch.sum(
-#                     (grid[:, :, :, None, None, :] - grid_ref[:, None, None, :, :, :])
-#                     ** 2,
-#                     dim=-1,
-#                 )
-#             )
-#             .reshape(batchsize, size_x, size_y, self.ref * self.ref)
-#             .contiguous()
-#         )
-#         return pos
-
-#     def forward(self, x, fx, T=None):
-#         with torch.autograd.profiler.record_function("prepeocess"):
-#             if self.unified_pos:
-#                 if self.pos.device != x.device:
-#                     # move the position tensor:
-#                     self.pos = self.pos.to(x.device)
-
-#                 x = (
-#                     self.pos.repeat(x.shape[0], 1, 1, 1)
-#                     .reshape(x.shape[0], self.H * self.W, self.ref * self.ref)
-#                     .to(x.device)
-#                 )
-#     if fx is not None:
-#         fx = torch.cat((x, fx), -1)
-#         fx = self.preprocess(fx)
-#     else:
-#         fx = self.preprocess(x)
-#         fx = fx + self.placeholder[None, None, :]
-
-# if T is not None:
-#     with torch.autograd.profiler.record_function("Time_Embedding"):
-#         Time_emb = timestep_embedding(T, self.n_hidden).repeat(1, x.shape[1], 1)
-#         Time_emb = self.time_fc(Time_emb)
-#         fx = fx + Time_emb
-
-# for i, block in enumerate(self.blocks):
-#     with torch.autograd.profiler.record_function(f"Block_{i}"):
-#         fx = block(fx)
-
-# return fx
-
-
 @dataclass
 class MetaData(ModelMetaData):
     name: str = "Transolver"
@@ -469,6 +322,7 @@ def __init__(
         super().__init__(meta=MetaData())
         self.__name__ = "Transolver"
 
+        self.use_te = use_te
         # Check that the hidden dimension and head dimensions are compatible:
         if not n_hidden % n_head == 0:
             raise ValueError(
@@ -550,6 +404,7 @@ def __init__(
             ]
         )
         self.initialize_weights()
+
         # self.placeholder = nn.Parameter(
         #     (1 / (n_hidden)) * torch.rand(n_hidden, dtype=torch.float)
         # )
@@ -562,7 +417,10 @@ def _init_weights(self, m):
             nn.init.trunc_normal_(m.weight, std=0.02)
             if isinstance(m, nn.Linear) and m.bias is not None:
                 nn.init.constant_(m.bias, 0)
-        elif isinstance(m, (nn.LayerNorm, te.LayerNorm, nn.BatchNorm1d)):
+        norm_layers = (nn.LayerNorm, nn.BatchNorm1d)
+        if self.use_te:
+            norm_layers = norm_layers + (te.LayerNorm,)
+        if isinstance(m, norm_layers):
             nn.init.constant_(m.bias, 0)
             nn.init.constant_(m.weight, 1.0)
 
@@ -627,7 +485,7 @@ def forward(
         Sublayers expect shape of [B, N, C]
 
         """
-        with torch.autograd.profiler.record_function("prepeocess"):
+        with torch.autograd.profiler.record_function("preprocess"):
 
             # Reshape automatically, if necessary:
             if self.structured_shape is not None:
@@ -646,10 +504,8 @@ def forward(
                     self.embedding.repeat(embedding.shape[0], 1, 1)
                     # .reshape(x.shape[0], -1, self.embedding_dim)
                 )
-
             # Combine the embedding and functional input:
             fx = torch.cat((embedding, fx), -1)
-
             # Dead code?  Not sure when this is used.
             # if fx is not None:
             # else:
@@ -676,113 +532,3 @@ def forward(
                 fx = fx.reshape(fx.shape[0], *self.structured_shape, -1)
 
         return fx
-
-
-# class Transolver(Module):
-#     """Transformer-based solver for PDEs.
-
-#     Note
-#     ----
-#     Transolver is a model specifically designed for structured 2D mesh data.
-
-#     Parameters
-#     ----------
-#     space_dim : int
-#         The spatial dimension of the input data.
-#     n_layers : int
-#         The number of transformer layers.
-#     n_hidden : int
-#         The hidden dimension of the transformer.
-#     dropout : float
-#         The dropout rate.
-#     n_head : int
-#         The number of attention heads.
-#     Time_Input : bool
-#         Whether to include time embeddings.
-#     act : str
-#         The activation function.
-#     mlp_ratio : int
-#         The ratio of hidden dimension in the MLP.
-#     fun_dim : int
-#         The dimension of the function.
-#     out_dim : int
-#         The output dimension.
-#     slice_num : int
-#         The number of slices in the structured attention.
-#     ref : int
-#         The reference dimension.
-#     unified_pos : bool
-#         Whether to use unified positional embeddings.
-#     structured_shape : None | tuple(int)
-#         The shape of the latent space.  If None, assumes irregulare latent space.
-#     use_te: bool
-#         Whether to use transformer engine backend when possible.
-#     """
-
-#     def __init__(
-#         self,
-#         space_dim: int,
-#         n_layers: int,
-#         n_hidden: int,
-#         dropout: float,
-#         n_head: int,
-#         Time_Input: bool,
-#         act: str,
-#         mlp_ratio: int,
-#         fun_dim: int,
-#         out_dim: int,
-#         slice_num: int,
-#         ref: int,
-#         unified_pos: bool,
-#         H: int,
-#         W: int,
-#         use_te: bool = True,
-#     ) -> None:
-#         super().__init__(meta=MetaData())
-#         self.H = H
-#         self.W = W
-#         self.model = Model(
-#             space_dim=space_dim,
-#             n_layers=n_layers,
-#             n_hidden=n_hidden,
-#             dropout=dropout,
-#             n_head=n_head,
-#             Time_Input=Time_Input,
-#             act=act,
-#             mlp_ratio=mlp_ratio,
-#             fun_dim=fun_dim,
-#             out_dim=out_dim,
-#             slice_num=slice_num,
-#             ref=ref,
-#             unified_pos=unified_pos,
-#             H=H,
-#             W=W,
-#             use_te=use_te,
-#         )
-
-#     def forward(
-#         self, x: torch.Tensor, fx: torch.Tensor = None, T: torch.Tensor = None
-#     ) -> torch.Tensor:
-#         """Forward pass.
-
-#         Parameters
-#         ----------
-#         x : torch.Tensor
-#             The input tensor.
-#         fx : torch.Tensor
-#             The function tensor.
-#         T : torch.Tensor
-#             The time tensor.
-
-#         Returns
-#         -------
-#         torch.Tensor
-#             The output tensor.
-
-#         """
-#         # print(f"Input x shape: {x.shape}")
-#         # print(f"Input fx shape: {fx.shape if fx is not None else None}")
-#         # print(f"Input T shape: {T.shape if T is not None else None}")
-#         y = self.model(x, fx, T)
-#         y = y.reshape(x.shape[0], self.H, self.W, -1)
-#         return y

From 554a3aba052491c86998a3b44a6863084880d8e4 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Thu, 17 Jul 2025 07:16:44 -0700
Subject: [PATCH 09/23] Enable volume training in transolver.  Still needs to
 be validated and optimized.

---
 examples/cfd/darcy_transolver/config_fix.yaml |  10 +-
 .../train_transolver_darcy_fix.py             | 119 ++++--
 .../transolver/README.md                      | 130 +++++++
 .../conf/{train.yaml => train_surface.yaml}   |  47 ++-
 .../transolver/conf/train_volume.yaml         |  93 +++++
 .../transolver/datapipe.py                    | 175 +++++++--
 .../external_aerodynamics/transolver/loss.py  | 162 +++++++-
 .../transolver/metrics.py                     |  46 ++-
 .../external_aerodynamics/transolver/train.py | 346 ++++++++++++++----
 9 files changed, 958 insertions(+), 170 deletions(-)
 rename examples/cfd/external_aerodynamics/transolver/conf/{train.yaml => train_surface.yaml} (73%)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml

diff --git a/examples/cfd/darcy_transolver/config_fix.yaml b/examples/cfd/darcy_transolver/config_fix.yaml
index e90fb50567..8f5cdcd2a0 100644
--- a/examples/cfd/darcy_transolver/config_fix.yaml
+++ b/examples/cfd/darcy_transolver/config_fix.yaml
@@ -21,12 +21,12 @@
 # SOFTWARE.
 
 output_dir: ./output/darcy_transolver_fix
-run_id: fp16_dev_fullres_b8_s64-te
+run_id: bf16_dev_r85_b8_s64
 
 data:
   train_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth1.npz
   test_path: /user_data/datasets/darcy_fix/example_data/piececonst_r421_N1024_smooth2.npz
-  resolution: 421 #421, 211, 141, 106, 85 all viable
+  resolution: 85 #421, 211, 141, 106, 85 all viable
   batch_size: 8 # This is the GLOBAL batch size
 
 model:
@@ -34,18 +34,18 @@ model:
   out_dim: 1
   embedding_dim: 2
   n_layers: 4
-  n_hidden: 64
+  n_hidden: 128
   dropout: 0.0
   n_head: 4
   act: gelu
-  mlp_ratio: 2
+  mlp_ratio: 4
   unified_pos: False
   ref: 8
   slice_num: 64
   use_te: True
   Time_Input: False
   
-precision: fp32
+precision: bf16
 
 normaliser:
   permeability:
diff --git a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
index 2b6461b470..e915778fab 100644
--- a/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
+++ b/examples/cfd/darcy_transolver/train_transolver_darcy_fix.py
@@ -53,7 +53,16 @@
 prof = Profiler()
 
 
-def count_parameters(model):
+def count_parameters(model: torch.nn.Module) -> int:
+    """
+    Count the total number of trainable parameters in a PyTorch model.
+
+    Args:
+        model (torch.nn.Module): The model whose parameters are to be counted.
+
+    Returns:
+        int: Total number of trainable parameters.
+    """
     total_params = 0
     for name, parameter in model.named_parameters():
         if not parameter.requires_grad:
@@ -64,14 +73,35 @@ def count_parameters(model):
 
 
 def forward_train_full_loop(
-    model, loss_fun, optimizer, pos, x, y, y_normalizer, context, scaler=None
-):
+    model: torch.nn.Module,
+    loss_fun: callable,
+    optimizer: torch.optim.Optimizer,
+    pos: torch.Tensor,
+    x: torch.Tensor,
+    y: torch.Tensor,
+    y_normalizer,
+    precision_context,
+    scaler: torch.cuda.amp.GradScaler = None,
+) -> torch.Tensor:
     """
     Forward and backward pass for one iteration, with optional mixed precision training.
-    Returns the loss for this minibatch
+
+    Args:
+        model (torch.nn.Module): The model to train.
+        loss_fun (callable): Loss function.
+        optimizer (torch.optim.Optimizer): Optimizer.
+        pos (torch.Tensor): Position tensor (embedding).
+        x (torch.Tensor): Input tensor.
+        y (torch.Tensor): Target tensor.
+        y_normalizer: Normalizer for the target tensor.
+        precision_context: Context manager for precision (e.g., autocast).
+        scaler (torch.cuda.amp.GradScaler, optional): GradScaler for mixed precision.
+
+    Returns:
+        torch.Tensor: The computed loss for this minibatch.
     """
     dm = DistributedManager()
-    with context:
+    with precision_context:
         pred = model(embedding=pos, fx=x.unsqueeze(-1)).squeeze(-1)
         pred = y_normalizer.decode(pred)
         loss = loss_fun(pred, y)
@@ -87,23 +117,43 @@ def forward_train_full_loop(
 
 
 def train_epoch(
-    model,
-    optimizer,
-    scheduler,
-    train_dataloader,
-    loss_fun,
+    model: torch.nn.Module,
+    optimizer: torch.optim.Optimizer,
+    scheduler: torch.optim.lr_scheduler._LRScheduler,
+    train_dataloader: DataLoader,
+    loss_fun: callable,
     y_normalizer,
-    context,
-    scaler,
-):
+    precision_context,
+    scaler: torch.cuda.amp.GradScaler,
+) -> torch.Tensor:
     """
-    One epoch of training.  Returns the loss from the last minibatch used, averaged across replicas.
-
+    One epoch of training. Returns the loss from the last minibatch used, averaged across replicas.
+
+    Args:
+        model (torch.nn.Module): The model to train.
+        optimizer (torch.optim.Optimizer): Optimizer.
+        scheduler (torch.optim.lr_scheduler._LRScheduler): Learning rate scheduler.
+        train_dataloader (DataLoader): Training data loader.
+        loss_fun (callable): Loss function.
+        y_normalizer: Normalizer for the target tensor.
+        precision_context: Context manager for precision (e.g., autocast).
+        scaler (torch.cuda.amp.GradScaler): GradScaler for mixed precision.
+
+    Returns:
+        torch.Tensor: The averaged loss from the last minibatch.
     """
     for i, batch in enumerate(train_dataloader):
         pos, x, y = batch
         loss = forward_train_full_loop(
-            model, loss_fun, optimizer, pos, x, y, y_normalizer, context, scaler
+            model,
+            loss_fun,
+            optimizer,
+            pos,
+            x,
+            y,
+            y_normalizer,
+            precision_context,
+            scaler,
         )
         scheduler.step()
 
@@ -116,9 +166,27 @@ def train_epoch(
     return loss
 
 
-def val_epoch(model, test_dataloader, loss_fun, y_normalizer):
+def val_epoch(
+    model: torch.nn.Module,
+    test_dataloader: DataLoader,
+    loss_fun: callable,
+    y_normalizer,
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
     """
-    One epoch of validation.  Returns the loss averaged across the entire validation set.
+    One epoch of validation. Returns the loss averaged across the entire validation set.
+
+    Args:
+        model (torch.nn.Module): The model to validate.
+        test_dataloader (DataLoader): Validation data loader.
+        loss_fun (callable): Loss function.
+        y_normalizer: Normalizer for the target tensor.
+
+    Returns:
+        tuple: (val_loss, pred, y, RL2)
+            val_loss (torch.Tensor): Averaged validation loss.
+            pred (torch.Tensor): Last batch predictions.
+            y (torch.Tensor): Last batch targets.
+            RL2 (torch.Tensor): Averaged relative L2 error.
     """
     val_loss = None
     RL2 = None
@@ -159,7 +227,12 @@ def val_epoch(model, test_dataloader, loss_fun, y_normalizer):
 
 @hydra.main(version_base="1.3", config_path=".", config_name="config_fix.yaml")
 def darcy_trainer(cfg: DictConfig) -> None:
-    """Training for the 2D Darcy flow benchmark problem."""
+    """
+    Training entry point for the 2D Darcy flow benchmark problem.
+
+    Args:
+        cfg (DictConfig): Configuration object loaded by Hydra.
+    """
     ########################################################################
     # Initialize distributed tools
     ########################################################################
@@ -301,13 +374,13 @@ def darcy_trainer(cfg: DictConfig) -> None:
 
     # Initialize GradScaler for mixed precision training
     if cfg.precision == "fp16":
-        context = torch.amp.autocast(device_type="cuda", dtype=torch.float16)
+        precision_context = torch.amp.autocast(device_type="cuda", dtype=torch.float16)
         scaler = torch.amp.GradScaler("cuda")
     elif cfg.precision == "bf16":
-        context = torch.amp.autocast(device_type="cuda", dtype=torch.bfloat16)
+        precision_context = torch.amp.autocast(device_type="cuda", dtype=torch.bfloat16)
         scaler = None
     else:
-        context = nullcontext()
+        precision_context = nullcontext()
         scaler = None
 
     if loaded_pseudo_epoch == 0:
@@ -331,7 +404,7 @@ def darcy_trainer(cfg: DictConfig) -> None:
                 train_dataloader,
                 loss_fun,
                 y_normalizer,
-                context,
+                precision_context,
                 scaler,
             )
             train_time = time.time() - train_start
diff --git a/examples/cfd/external_aerodynamics/transolver/README.md b/examples/cfd/external_aerodynamics/transolver/README.md
index e69de29bb2..363cc5f037 100644
--- a/examples/cfd/external_aerodynamics/transolver/README.md
+++ b/examples/cfd/external_aerodynamics/transolver/README.md
@@ -0,0 +1,130 @@
+# Transolver CFD Example: Code Overview
+
+This directory contains the core components for training and evaluating a Transolver
+model for external aerodynamics CFD problems.  The Transolver model is an adaptation
+of the Attention mechanism to encourage models to learn meaningful representations.
+In each PhysicsAttention layer, the input points are projected onto state vectors via
+learnable transformations and weights.  These learnable transformations are in turn
+used to calculate self-attention between all state vectors, and the weights are reused
+to project states back to each input point.
+
+Through a series of PhysicsAttention layers, the Transolver model learns high quality
+projections from functional input space to output space.  The PhysicsNeMo implementation
+of Transolver is an equivalent implementation of the original model architecture
+([https://github.com/thuml/Transolver](https://github.com/thuml/Transolver)), with
+modifications to improve numerical stability and support NVIDIA TransformerEngine.
+
+The Transolver training recipe reuses the same input datasets as DoMINO - please see
+the external_aerodynamics example for DoMINO for more information, as well as
+[PhysicsNeMo Curator](https://github.com/NVIDIA/physicsnemo-curator) for information
+on producing the data if needed.  The Transolver model does not require the neighbor
+points as input to the model, nor does it require any structure to the input points
+(graph connections, edges, etc.).  Consequently, the datapipe is significantly simpler
+than for DoMINO - the entire datapipe is encapulated here in the examples.
+
+Below is a high-level overview of the main files and their roles in the workflow.
+
+---
+
+## 1. `conf/train.yaml`
+
+**Purpose:**  
+Configuration file for training runs.  
+**Contents:**  
+
+- **Output and run settings:** Output directory, run ID, random seed, and precision.
+- **Training parameters:** Number of epochs, checkpoint intervals, and whether to use
+  compilation.
+- **Model configuration:** Architecture details (input/output dimensions, number of
+  layers, embedding size, attention heads, activation, etc.).
+- **Optimizer settings:** Type (AdamW), learning rate, weight decay, and optimizer
+  hyperparameters.
+- **Data settings:** Paths to training/validation datasets, worker/thread settings,
+  memory pinning, and which data keys to load.
+- **Logging:** Logging level and format.
+
+---
+
+## 2. `src/train.py`
+
+**Purpose:**  
+Main training script for the DoMINO/Transolver model using distributed data parallelism.
+
+**Key Features:**
+
+- **Distributed Training:** Uses PyTorch DistributedDataParallel for multi-GPU training.
+- **Data Loading:** Loads datasets using custom datapipe, supports distributed sampling.
+- **Model Instantiation:** Builds the model based on config, supports both surface and
+  volume predictions.
+- **Loss Calculation:** Computes losses for volume, surface, and integral quantities
+  (lift/drag).
+- **Mixed Precision:** Supports mixed-precision training with gradient scaling.
+- **Checkpointing:** Automatically loads/saves checkpoints and tracks best validation
+  loss.
+- **Logging:** Logs metrics to TensorBoard and console, including GPU memory usage.
+- **Validation:** Evaluates model on validation set after each epoch.
+
+---
+
+## 3. `loss.py`
+
+**Purpose:**  
+Defines loss functions for training the Transolver model.
+
+**Key Features:**
+
+- **Surface Loss:** Computes MSE or RMSE for both pressure (scalar) and wall shear
+  (vector) components, handling them separately and combining the results.
+- **Integral Losses:** Implements physics-based losses for lift and drag, using surface
+  integrals of predicted and true values, weighted by area, normals, and stream
+  velocity.
+- **Modularity:** Loss functions are modular and can be combined or extended for
+  different training objectives.
+
+---
+
+## 4. `metrics.py`
+
+**Purpose:**  
+Defines evaluation metrics for model predictions.
+
+**Key Features:**
+
+- **Distributed Reduction:** Aggregates metrics across distributed processes.
+- **Surface Metrics:** Computes normalized L2 errors for pressure and shear components,
+  after unnormalizing predictions and targets.
+- **Extensibility:** Can be extended for additional metrics or domains.
+
+---
+
+## 5. `datapipe.py`
+
+**Purpose:**  
+Implements a PyTorch dataset for efficient loading of large CFD datasets stored in Zarr
+format.
+
+**Key Features:**
+
+- **Chunk-Aligned I/O:** Reads large arrays in chunk-aligned fashion for efficiency,
+  using threads for parallel I/O.
+- **Flexible Key Loading:** Allows specifying which data keys to load and which are
+  considered "large" (for chunked reading).
+- **Pinned Memory:** Optionally allocates pinned memory for faster GPU transfers.
+- **Prefetching:** Supports asynchronous preloading of samples to overlap I/O and
+  computation.
+
+---
+
+## Summary
+
+- **`train.yaml`**: All configuration for model, data, optimizer, and logging.
+- **`train.py`**: Orchestrates distributed training, validation, checkpointing, and
+  logging.
+- **`loss.py`**: Physics-informed and standard loss functions for model training.
+- **`metrics.py`**: Evaluation metrics for model performance, with distributed support.
+- **`datapipe.py`**: High-performance, domain-parallel data loading from Zarr files for
+  large-scale CFD datasets.
+
+---
+
+For more details, refer to the docstrings and comments within each file.
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
similarity index 73%
rename from examples/cfd/external_aerodynamics/transolver/conf/train.yaml
rename to examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
index 37dc470d9e..db56902f48 100644
--- a/examples/cfd/external_aerodynamics/transolver/conf/train.yaml
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
@@ -16,15 +16,17 @@
 
 
 
-output_dir: "outputs"
-run_id: "transolver_te-1cycle"
+output_dir: "runs"
+run_id: "surface/ReduceLR/No-TE/float16/"
 
 # Training configuration
 training:
-  num_epochs: 400
+  precision: float16 # float32, float16, bfloat16
+  num_epochs: 500
   save_interval: 25  # Save checkpoint every N epochs
   seed: 42
-  output_dir: "outputs"
+  output_dir: "runs"
+  compile: false
 
 # Model configuration
 model:
@@ -42,15 +44,24 @@ model:
   unified_pos: false # Whether to use unified positional embeddings
   ref: 8            # Reference dimension for unified pos
   structured_shape: null
-  use_te: true      # Use transformer engine
+  use_te: false      # Use transformer engine
   Time_Input: false # Whether to use time embeddings
 
+scheduler:
+  name: "OneCycleLR"
+  params:
+    final_div_factor: 1e4
+
+# scheduler:
+#   name: "ReduceLROnPlateau"
+#   params:
+#     min_lr: 1e-6
 
 
 # Optimizer configuration
 optimizer:
   _target_: torch.optim.AdamW
-  lr: 1.0e-3
+  lr: 3.0e-4
   weight_decay: 1.0e-4
   betas: [0.9, 0.999]
   eps: 1.0e-8
@@ -58,25 +69,31 @@ optimizer:
 # Data configuration
 data:
   train:
-    data_path: /group_data/datasets/drivaer_aws/domino/train/
+    data_path: /user_data/datasets/domino/train/
   val:
-    data_path: /group_data/datasets/drivaer_aws/domino/val/
-  max_workers: 32
+    data_path: /user_data/datasets/domino/val/
+  max_workers: 64
   pin_memory: true
-  resolution: 262_144
-  surface_keys:
+  # resolution: 262_144
+  resolution: 325_000
+  mode: surface
+  data_keys:
     - "surface_fields"
     - "surface_mesh_centers"
     - "surface_normals"
     - "air_density"
     - "stream_velocity"
+    - "stl_areas"
+    - "stl_centers"
+  large_keys:
+    - "surface_fields"
+    - "surface_mesh_centers"
+    - "surface_normals"
+    - "stl_areas"
+    - "stl_centers"
 
 # Logging configuration
 logging:
   level: INFO
   format: '%(asctime)s - %(name)s - %(levelname)s - %(message)s'
 
-# Hardware configuration
-hardware:
-  precision: float32
-  compile_model: false  # Whether to use torch.compile
\ No newline at end of file
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
new file mode 100644
index 0000000000..889a57c0ff
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
@@ -0,0 +1,93 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+
+output_dir: "runs"
+run_id: "volume/OneCycleLR/TE/float32/"
+
+# Training configuration
+training:
+  precision: float32 # float32, float16, bfloat16
+  num_epochs: 200
+  save_interval: 25  # Save checkpoint every N epochs
+  seed: 42
+  output_dir: "runs"
+  compile: false
+
+# Model configuration
+model:
+  _target_: physicsnemo.models.transolver.Transolver
+  functional_dim: 2 # Input feature dimension
+  out_dim: 5        # Output feature dimension
+  embedding_dim: 3  # Spatial embedding dimension
+  n_layers: 8       # Number of transformer layers
+  n_hidden: 256     # Hidden dimension
+  dropout: 0.0      # Dropout rate
+  n_head: 8         # Number of attention heads
+  act: "gelu"       # Activation function
+  mlp_ratio: 4      # MLP ratio in attention blocks
+  slice_num: 128     # Number of slices in physics attention
+  unified_pos: false # Whether to use unified positional embeddings
+  ref: 8            # Reference dimension for unified pos
+  structured_shape: null
+  use_te: true      # Use transformer engine
+  Time_Input: false # Whether to use time embeddings
+
+scheduler:
+  name: "OneCycleLR"
+  params:
+    final_div_factor: 1e4
+
+# scheduler:
+#   name: "ReduceLROnPlateau"
+#   params:
+#     min_lr: 1e-6
+
+
+# Optimizer configuration
+optimizer:
+  _target_: torch.optim.AdamW
+  lr: 1.0e-3
+  weight_decay: 1.0e-4
+  betas: [0.9, 0.999]
+  eps: 1.0e-8
+
+# Data configuration
+data:
+  train:
+    data_path: /user_data/datasets/domino/train/
+  val:
+    data_path: /user_data/datasets/domino/val/
+  max_workers: 64
+  pin_memory: true
+  # resolution: 262_144
+  resolution: 200_000
+  mode: volume
+  data_keys:
+    - "volume_mesh_centers"
+    - "volume_fields"
+    - "air_density"
+    - "stream_velocity"
+  large_keys:
+    - "volume_fields"
+    - "volume_mesh_centers"
+
+# Logging configuration
+logging:
+  level: INFO
+  format: '%(asctime)s - %(name)s - %(levelname)s - %(message)s'
+
diff --git a/examples/cfd/external_aerodynamics/transolver/datapipe.py b/examples/cfd/external_aerodynamics/transolver/datapipe.py
index c1de86b431..1c9e168c01 100644
--- a/examples/cfd/external_aerodynamics/transolver/datapipe.py
+++ b/examples/cfd/external_aerodynamics/transolver/datapipe.py
@@ -54,14 +54,21 @@
 from physicsnemo.distributed.utils import compute_split_shapes
 from physicsnemo.utils.profiling import profile
 
+import threading
+
 
 def get_filenames(data_path: Path, exclude_dirs: bool = True) -> List[str]:
-    """Get list of filenames from data directory."""
+    """Get list of filenames from data directory.
+
+    Args:
+        data_path: Path to the directory containing files.
+        exclude_dirs: If True, exclude directories from the result.
 
+    Returns:
+        Sorted list of filenames with .zarr extension.
+    """
     filenames = []
     for item in data_path.iterdir():
-        # if exclude_dirs and item.is_dir():
-        #     continue
         if item.suffix in [".zarr"]:
             filenames.append(item.name)
     return sorted(filenames)
@@ -72,8 +79,15 @@ def _read_chunk_into_array(
     zarr_array: zarr.Array,
     cpu_slice: slice,
     zarr_slice: slice = None,
-):
-    """Helper function to read a chunk from zarr into numpy array."""
+) -> None:
+    """Helper function to read a chunk from zarr into numpy array.
+
+    Args:
+        cpu_array: The destination numpy array.
+        zarr_array: The source zarr array.
+        cpu_slice: The slice in the cpu_array to write to.
+        zarr_slice: The slice in the zarr_array to read from. If None, uses cpu_slice.
+    """
     if zarr_slice is None:
         zarr_slice = cpu_slice
     cpu_array[cpu_slice] = zarr_array[zarr_slice]
@@ -81,7 +95,14 @@ def _read_chunk_into_array(
 
 @lru_cache
 def to_torch_dtype(dtype: np.dtype) -> torch.dtype:
-    """Convert a numpy dtype to a torch dtype."""
+    """Convert a numpy dtype to a torch dtype.
+
+    Args:
+        dtype: Numpy dtype to convert.
+
+    Returns:
+        Corresponding torch dtype.
+    """
     temp = torch.from_numpy(np.empty((), dtype=dtype))
     return temp.dtype
 
@@ -199,6 +220,7 @@ def __init__(
         self.tensor_specs = {}
 
     def __len__(self) -> int:
+        """Return the number of samples in the dataset."""
         return len(self.filenames)
 
     def __del__(self):
@@ -213,6 +235,14 @@ def _read_key_chunk_aligned(
         """
         Read a key using chunk-aligned I/O for efficiency.
         Implements domain parallelism if device_mesh is configured.
+
+        Args:
+            zarr_group: The zarr group to read from.
+            key: The key to read.
+            futures: List to append thread futures to.
+
+        Returns:
+            Torch tensor containing the read data.
         """
 
         zarr_array = zarr_group[key]
@@ -325,7 +355,16 @@ def _read_key_chunk_aligned(
     def _read_key_standard(
         self, zarr_group: zarr.Group, key: str, futures: list[Future]
     ) -> torch.Tensor:
-        """Read a key with simple I/O (for small arrays)."""
+        """Read a key with simple I/O (for small arrays).
+
+        Args:
+            zarr_group: The zarr group to read from.
+            key: The key to read.
+            futures: List to append thread futures to.
+
+        Returns:
+            Torch tensor containing the read data.
+        """
         zarr_array = zarr_group[key]
 
         # Handle scalar values
@@ -376,7 +415,14 @@ def _read_key_standard(
 
     @profile
     def _read_zarr_file(self, filepath: Path) -> Dict[str, np.ndarray]:
-        """Read data from a zarr file."""
+        """Read data from a zarr file.
+
+        Args:
+            filepath: Path to the zarr file.
+
+        Returns:
+            Dictionary mapping keys to numpy arrays or torch tensors.
+        """
         with zarr.open_group(filepath, mode="r") as zarr_group:
             data = {}
             futures = []
@@ -402,7 +448,14 @@ def _read_zarr_file(self, filepath: Path) -> Dict[str, np.ndarray]:
 
     @profile
     def _move_to_gpu(self, data: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]:
-        """Convert numpy arrays to torch tensors and move to GPU if available."""
+        """Convert numpy arrays to torch tensors and move to GPU if available.
+
+        Args:
+            data: Dictionary of key to torch tensor.
+
+        Returns:
+            Dictionary of key to torch tensor on GPU if available.
+        """
         result = {}
 
         for key, array in data.items():
@@ -415,7 +468,14 @@ def _move_to_gpu(self, data: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]
     def _convert_to_shard_tensors(
         self, tensors: Dict[str, torch.Tensor]
     ) -> Dict[str, Union[torch.Tensor, ShardTensor]]:
-        """Convert tensors to ShardTensor objects for distributed training."""
+        """Convert tensors to ShardTensor objects for distributed training.
+
+        Args:
+            tensors: Dictionary of key to torch tensor.
+
+        Returns:
+            Dictionary of key to torch tensor or ShardTensor.
+        """
         if self.device_mesh is None:
             return tensors
 
@@ -446,6 +506,55 @@ def _convert_to_shard_tensors(
 
         return result
 
+    def preload(self, idx: int) -> None:
+        """
+        Asynchronously preload the data for the given index (up to CPU, not GPU).
+        Only one preload operation is supported at a time.
+
+        Args:
+            idx: Index of the sample to preload.
+        """
+        if hasattr(self, "_preload_thread") and self._preload_thread is not None:
+            # Optionally, wait for previous preload to finish or raise error
+            self._preload_thread.join()
+
+        self._preload_result = None
+        self._preload_exception = None
+
+        def _preload_worker():
+            try:
+                filename = self.filenames[idx]
+                filepath = self.data_path / filename
+                data = self._read_zarr_file(filepath)
+                self._preload_result = (idx, data)
+            except Exception as e:
+                self._preload_exception = e
+
+        self._preload_thread = threading.Thread(target=_preload_worker)
+        self._preload_thread.start()
+
+    def get_preloaded(self) -> Tuple[int, Dict[str, np.ndarray]]:
+        """
+        Retrieve the preloaded data (blocking if not ready).
+
+        Returns:
+            (idx, data) tuple where data is a dictionary of key to numpy array or torch tensor.
+
+        Raises:
+            RuntimeError: If no preload is in progress.
+            Exception: If preload failed.
+        """
+        if not hasattr(self, "_preload_thread") or self._preload_thread is None:
+            raise RuntimeError("No preload in progress. Call preload(idx) first.")
+
+        self._preload_thread.join()
+        self._preload_thread = None
+
+        if self._preload_exception is not None:
+            raise self._preload_exception
+
+        return self._preload_result
+
     @profile
     def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
         """
@@ -462,11 +571,26 @@ def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
                 f"Index {idx} out of range for dataset of size {len(self.filenames)}"
             )
 
-        filename = self.filenames[idx]
-        filepath = self.data_path / filename
+        if hasattr(self, "_preload_result") and self._preload_result is not None:
+            preload_idx, preload_data = self._preload_result
+            if preload_idx == idx:
+                data = preload_data
+                self._preload_result = None  # Clear after use
+            else:
+                # Preloaded data is for a different idx, ignore it
+                data = self._read_zarr_file(self.data_path / self.filenames[idx])
+        else:
+            filename = self.filenames[idx]
+            filepath = self.data_path / filename
+
+            # Read data from zarr file
+            data = self._read_zarr_file(filepath)
 
-        # Read data from zarr file
-        data = self._read_zarr_file(filepath)
+        # filename = self.filenames[idx]
+        # filepath = self.data_path / filename
+
+        # # Read data from zarr file
+        # data = self._read_zarr_file(filepath)
 
         # Convert to torch tensors
         tensors = self._move_to_gpu(data)
@@ -485,26 +609,3 @@ def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
 # 4. Support for different placement strategies per key
 # 5. Memory-mapped reading for very large datasets
 # 6. Compression/decompression handling
-
-
-def create_transolver_dataset(
-    data_path: Union[str, Path],
-    device_mesh: Optional[DeviceMesh] = None,
-    placements: Optional[Tuple[Placement, ...]] = None,
-    **kwargs,
-) -> DomainParallelZarrDataset:
-    """
-    Factory function to create a Transolver dataset with sensible defaults.
-
-    Args:
-        data_path: Path to zarr data directory
-        device_mesh: Optional device mesh for domain parallelism
-        placements: Optional placement specifications
-        **kwargs: Additional arguments passed to DomainParallelZarrDataset
-
-    Returns:
-        Configured dataset instance
-    """
-    return DomainParallelZarrDataset(
-        data_path=data_path, device_mesh=device_mesh, placements=placements, **kwargs
-    )
diff --git a/examples/cfd/external_aerodynamics/transolver/loss.py b/examples/cfd/external_aerodynamics/transolver/loss.py
index 3420d7324c..37dd31d969 100644
--- a/examples/cfd/external_aerodynamics/transolver/loss.py
+++ b/examples/cfd/external_aerodynamics/transolver/loss.py
@@ -18,8 +18,50 @@
 from typing import Literal
 
 
-def loss_fn(pred, target):
-    return loss_fn_surface(pred, target, "mse")
+def loss_fn(
+    pred: torch.Tensor,
+    target: torch.Tensor,
+    others: dict[str, torch.Tensor],
+    mode: Literal["surface", "volume"],
+) -> torch.Tensor:
+    """
+    Compute the main loss function for the model.
+
+    Args:
+        pred: Predicted tensor from the model.
+        target: Ground truth tensor.
+        others: Dictionary of additional tensors (e.g., surface_areas, surface_normals, stream_velocity).
+
+    Returns:
+        Loss value as a scalar tensor.
+    """
+    if mode == "surface":
+        loss = loss_fn_surface(pred, target, "mse")
+    elif mode == "volume":
+        loss = loss_fn_volume(pred, target, "rmse")
+    # 100 * integral_loss_fn(pred, target, others["surface_areas"], others["surface_normals"], others["stream_velocity"])
+    return loss
+
+
+def loss_fn_volume(
+    pred: torch.Tensor, target: torch.Tensor, mode: Literal["mse", "rmse"]
+) -> torch.Tensor:
+    """
+    Compute the main loss function for the model.
+    """
+
+    if mode == "rmse":
+        dims = (0, 1)
+    else:
+        dims = None
+
+    num = torch.sum((pred - target) ** 2.0, dims)
+    if mode == "rmse":
+        denom = torch.sum(target**2.0, dims)
+    else:
+        denom = pred.shape[1]
+
+    return torch.mean(num / denom)
 
 
 def loss_fn_surface(
@@ -28,12 +70,12 @@ def loss_fn_surface(
     """Calculate loss for surface data by handling scalar and vector components separately.
 
     Args:
-        output: Predicted surface values from the model
-        target: Ground truth surface values
-        loss_type: Type of loss to calculate ("mse" or "rmse")
+        output: Predicted surface values from the model.
+        target: Ground truth surface values.
+        loss_type: Type of loss to calculate ("mse" or "rmse").
 
     Returns:
-        Combined scalar and vector loss as a scalar tensor
+        Combined scalar and vector loss as a scalar tensor.
     """
     # Separate the scalar and vector components:
     output_pressure, output_sheer = torch.split(output, [1, 3], dim=2)
@@ -56,3 +98,111 @@ def loss_fn_surface(
     loss = loss_pressure + loss_wall_sheer
 
     return loss / 4.0
+
+
+def integral_loss_fn(
+    output: torch.Tensor,
+    target: torch.Tensor,
+    area: torch.Tensor,
+    normals: torch.Tensor,
+    stream_velocity: torch.Tensor = None,
+) -> torch.Tensor:
+    """
+    Compute the integral loss (sum of lift and drag losses).
+
+    Args:
+        output: Predicted tensor.
+        target: Ground truth tensor.
+        area: Surface area tensor.
+        normals: Surface normals tensor.
+        stream_velocity: Stream velocity tensor (optional).
+
+    Returns:
+        Scalar tensor representing the sum of lift and drag losses.
+    """
+    drag_loss = drag_loss_fn(output, target, area, normals, stream_velocity)
+    lift_loss = lift_loss_fn(output, target, area, normals, stream_velocity)
+    return lift_loss + drag_loss
+
+
+def lift_loss_fn(
+    output: torch.Tensor,
+    target: torch.Tensor,
+    area: torch.Tensor,
+    normals: torch.Tensor,
+    stream_velocity: torch.Tensor = None,
+) -> torch.Tensor:
+    """
+    Compute the lift loss.
+
+    Args:
+        output: Predicted tensor.
+        target: Ground truth tensor.
+        area: Surface area tensor.
+        normals: Surface normals tensor.
+        stream_velocity: Stream velocity tensor (optional).
+
+    Returns:
+        Scalar tensor representing the lift loss.
+    """
+    vel_inlet = stream_velocity  # Get this from the dataset
+    # mask = abs(target - padded_value) > 1e-3
+
+    output_true = target * area * (vel_inlet) ** 2.0
+    output_pred = output * area * (vel_inlet) ** 2.0
+
+    normals = torch.select(normals, 2, 2)
+    output_true_0 = output_true.select(2, 0)
+    output_pred_0 = output_pred.select(2, 0)
+
+    pres_true = output_true_0 * normals
+    pres_pred = output_pred_0 * normals
+
+    wz_true = output_true[:, :, -1]
+    wz_pred = output_pred[:, :, -1]
+
+    masked_pred = torch.mean(pres_pred + wz_pred, (1))
+    masked_truth = torch.mean(pres_true + wz_true, (1))
+
+    loss = (masked_pred - masked_truth) ** 2.0
+    loss = torch.mean(loss)
+    return loss
+
+
+def drag_loss_fn(
+    output: torch.Tensor,
+    target: torch.Tensor,
+    area: torch.Tensor,
+    normals: torch.Tensor,
+    stream_velocity: torch.Tensor = None,
+) -> torch.Tensor:
+    """
+    Compute the drag loss.
+
+    Args:
+        output: Predicted tensor.
+        target: Ground truth tensor.
+        area: Surface area tensor.
+        normals: Surface normals tensor.
+        stream_velocity: Stream velocity tensor (optional).
+
+    Returns:
+        Scalar tensor representing the drag loss.
+    """
+    vel_inlet = stream_velocity  # Get this from the dataset
+    # mask = abs(target - padded_value) > 1e-3
+    output_true = target * area * (vel_inlet) ** 2.0
+    output_pred = output * area * (vel_inlet) ** 2.0
+
+    pres_true = output_true[:, :, 0] * normals[:, :, 0]
+    pres_pred = output_pred[:, :, 0] * normals[:, :, 0]
+
+    wx_true = output_true[:, :, 1]
+    wx_pred = output_pred[:, :, 1]
+
+    masked_pred = torch.mean(pres_pred + wx_pred, (1))
+    masked_truth = torch.mean(pres_true + wx_true, (1))
+
+    loss = (masked_pred - masked_truth) ** 2.0
+    loss = torch.mean(loss)
+    return loss
diff --git a/examples/cfd/external_aerodynamics/transolver/metrics.py b/examples/cfd/external_aerodynamics/transolver/metrics.py
index 5bba878e2a..2b3dfc1939 100644
--- a/examples/cfd/external_aerodynamics/transolver/metrics.py
+++ b/examples/cfd/external_aerodynamics/transolver/metrics.py
@@ -34,11 +34,45 @@ def all_reduce_dict(metrics, dm):
     return metrics
 
 
-def metrics_fn(pred, target, dm):
-    return metrics_fn_surface(pred, target, dm)
+def metrics_fn(pred, target, others, dm, mode):
+    if mode == "surface":
+        return metrics_fn_surface(pred, target, others, dm)
+    elif mode == "volume":
+        return metrics_fn_volume(pred, target, others, dm)
+    else:
+        raise ValueError(f"Unknown data mode: {mode}")
 
 
-def metrics_fn_surface(pred, target, dm):
+def metrics_fn_volume(pred, target, others, dm):
+    target = target * others["norm_std"] + others["norm_mean"]
+    pred = pred * others["norm_std"] + others["norm_mean"]
+
+    l2_num = (pred - target) ** 2
+    l2_num = torch.sum(l2_num, dim=1)
+    l2_num = torch.sqrt(l2_num)
+
+    l2_denom = target**2
+    l2_denom = torch.sum(l2_denom, dim=1)
+    l2_denom = torch.sqrt(l2_denom)
+
+    l2 = l2_num / l2_denom
+
+    metrics = {
+        "l2_volume_1": torch.mean(l2[:, 0]),
+        "l2_volume_2": torch.mean(l2[:, 1]),
+        "l2_volume_3": torch.mean(l2[:, 2]),
+        "l2_volume_4": torch.mean(l2[:, 3]),
+        "l2_volume_5": torch.mean(l2[:, 4]),
+    }
+
+    return metrics
+
+
+def metrics_fn_surface(pred, target, others, dm):
+
+    # Unnormalize the surface values for L2:
+    target = target * others["norm_std"] + others["norm_mean"]
+    pred = pred * others["norm_std"] + others["norm_mean"]
 
     l2_num = (pred - target) ** 2
     l2_num = torch.sum(l2_num, dim=1)
@@ -52,9 +86,9 @@ def metrics_fn_surface(pred, target, dm):
 
     metrics = {
         "l2_pressure": torch.mean(l2[:, 0]),
-        "l2_sheer_x": torch.mean(l2[:, 1]),
-        "l2_sheer_y": torch.mean(l2[:, 2]),
-        "l2_sheer_z": torch.mean(l2[:, 3]),
+        "l2_shear_x": torch.mean(l2[:, 1]),
+        "l2_shear_y": torch.mean(l2[:, 2]),
+        "l2_shear_z": torch.mean(l2[:, 3]),
     }
 
     return metrics
diff --git a/examples/cfd/external_aerodynamics/transolver/train.py b/examples/cfd/external_aerodynamics/transolver/train.py
index 12b2e78708..26f348f04d 100644
--- a/examples/cfd/external_aerodynamics/transolver/train.py
+++ b/examples/cfd/external_aerodynamics/transolver/train.py
@@ -34,9 +34,60 @@
 from loss import loss_fn
 from metrics import metrics_fn
 
+from contextlib import nullcontext
+from torch.amp import autocast, GradScaler
+
+# import transformer_engine.pytorch as te
+# from transformer_engine.common.recipe import Format, DelayedScaling
+
+
+def get_autocast_context(precision: str) -> nullcontext:
+    """
+    Returns the appropriate autocast context for mixed precision training.
+
+    Args:
+        precision (str): The desired precision. Supported values are "float16", "bfloat16", or any other string for no autocast.
+
+    Returns:
+        Context manager: An autocast context for the specified precision, or a nullcontext if precision is not recognized.
+    """
+    if precision == "float16":
+        return autocast("cuda", dtype=torch.float16)
+    elif precision == "bfloat16":
+        return autocast("cuda", dtype=torch.bfloat16)
+    # elif precision == "float8":
+    #     print("Using float8 autocast")
+    #     fp8_format = Format.HYBRID
+    #     fp8_recipe = DelayedScaling(fp8_format=fp8_format, amax_history_len=16, amax_compute_algo="max")
+    #     return te.fp8_autocast(enabled=True, fp8_recipe=fp8_recipe)
+    else:
+        return nullcontext()
+
+
+def cast_precisions(
+    features: torch.Tensor, embeddings: torch.Tensor, precision: str
+) -> tuple[torch.Tensor, torch.Tensor]:
+    """
+    Casts the features and embeddings tensors to the specified precision.
+
+    Args:
+        features (torch.Tensor): The input features tensor.
+        embeddings (torch.Tensor): The input embeddings tensor.
+        precision (str): The desired precision ("float16", "bfloat16", or other for no cast).
+
+    Returns:
+        Tuple[torch.Tensor, torch.Tensor]: The features and embeddings tensors cast to the specified precision.
+    """
+    if precision == "float16":
+        return features.to(torch.float16), embeddings.to(torch.float16)
+    elif precision == "bfloat16":
+        return features.to(torch.bfloat16), embeddings.to(torch.bfloat16)
+    else:
+        return features, embeddings
+
 
 @profile
-def preprocess_data(batch):
+def preprocess_surface_data(batch):
 
     mesh_centers = batch["surface_mesh_centers"]
     normals = batch["surface_normals"]
@@ -45,6 +96,10 @@ def preprocess_data(batch):
         [batch["air_density"], batch["stream_velocity"]], dim=-1
     ).to(torch.float32)
 
+    # Normalize the surface fields:
+    norm_mean = targets.mean(dim=0)
+    norm_std = targets.std(dim=0)
+    targets = (targets - norm_mean) / norm_std
     # fourier_sin_features = [
     #     torch.sin(mesh_centers * (2 ** i) * torch.pi)
     #     for i in range(4)
@@ -54,6 +109,17 @@ def preprocess_data(batch):
     #     for i in range(4)
     # ]
 
+    # Calculate center of mass
+    sizes = batch["stl_areas"]
+    centers = batch["stl_centers"]
+
+    total_weighted_position = torch.einsum("i,ij->j", sizes, centers)
+    total_size = torch.sum(sizes)
+    center_of_mass = total_weighted_position[None, ...] / total_size
+
+    # Subtract the COM from the centers:
+    mesh_centers = mesh_centers - center_of_mass
+
     embeddings = torch.cat(
         [
             mesh_centers,
@@ -66,16 +132,70 @@ def preprocess_data(batch):
 
     node_features = node_features.unsqueeze(0).broadcast_to(embeddings.shape[0], -1)
 
-    return node_features, embeddings, targets
+    others = {
+        "surface_areas": sizes,
+        "surface_normals": normals,
+        "stream_velocity": batch["stream_velocity"],
+        "air_density": batch["air_density"],
+        "norm_mean": norm_mean,
+        "norm_std": norm_std,
+    }
+
+    return node_features, embeddings, targets, others
+
+
+def preprocess_volume_data(batch):
+
+    mesh_centers = batch["volume_mesh_centers"]
+    targets = batch["volume_fields"]
+    node_features = torch.stack(
+        [batch["air_density"], batch["stream_velocity"]], dim=-1
+    ).to(torch.float32)
+
+    embeddings = mesh_centers
+
+    # Normalize the surface fields:
+    norm_mean = targets.mean(dim=0)
+    norm_std = targets.std(dim=0)
+    targets = (targets - norm_mean) / norm_std
+
+    node_features = node_features.unsqueeze(0).broadcast_to(embeddings.shape[0], -1)
+
+    others = {
+        "norm_mean": norm_mean,
+        "norm_std": norm_std,
+    }
+    return node_features, embeddings, targets, others
 
 
 @profile
-def downsample(features, embeddings, targets, num_keep=1024):
+def downsample_surface(features, embeddings, targets, num_keep=1024):
     # Determine the number of samples to keep (e.g., 50% of original size)
     num_samples = features.shape[0]
+    # Generate random indices to keep (faster for large num_samples)
+    indices = torch.multinomial(
+        torch.ones(num_samples, device=features.device), num_keep, replacement=False
+    )
 
-    # Generate random indices to keep
-    indices = torch.randperm(num_samples)[:num_keep]
+    # Use the same indices to downsample all tensors
+    downsampled_features = features[indices]
+    downsampled_embeddings = embeddings[indices]
+    downsampled_targets = targets[indices]
+
+    downsampled_features = downsampled_features.unsqueeze(0)
+    downsampled_embeddings = downsampled_embeddings.unsqueeze(0)
+    downsampled_targets = downsampled_targets.unsqueeze(0)
+
+    return downsampled_features, downsampled_embeddings, downsampled_targets
+
+
+@profile
+def downsample_volume(features, embeddings, targets, num_keep=1024):
+    # Determine the number of samples to keep (e.g., 50% of original size)
+    num_samples = features.shape[0]
+    # The volume data is so large, that we'll sample randints
+    # which will very rarely duplicate
+    indices = torch.randint(0, num_samples, (num_keep,), device=features.device)
 
     # Use the same indices to downsample all tensors
     downsampled_features = features[indices]
@@ -101,6 +221,7 @@ def train_epoch(
     epoch,
     cfg,
     dist_manager,
+    scaler=None,  # <-- Added scaler argument
 ):
     """Train for one epoch
 
@@ -117,63 +238,84 @@ def train_epoch(
 
     epoch_indices = list(sampler)
     epoch_len = len(epoch_indices)
-
+    precision = getattr(cfg.training, "precision", "float32")
+    context = get_autocast_context(precision)
     start_time = time.time()
-    for i, batch_idx in enumerate(epoch_indices):
-        batch = dataloader[batch_idx]
-        # Get data from batch
-        features, embeddings, targets = preprocess_data(batch)
+    with Profiler():
+        for i, batch_idx in enumerate(epoch_indices):
+            batch = dataloader[batch_idx]
+            # Get data from batch
+            if cfg.data.mode == "surface":
+                features, embeddings, targets, others = preprocess_surface_data(batch)
+                features, embeddings, targets = downsample_surface(
+                    features, embeddings, targets, cfg.data.resolution
+                )
+            elif cfg.data.mode == "volume":
+                features, embeddings, targets, others = preprocess_volume_data(batch)
+                features, embeddings, targets = downsample_volume(
+                    features, embeddings, targets, cfg.data.resolution
+                )
+            else:
+                raise ValueError(f"Unknown data mode: {cfg.data.mode}")
 
-        features, embeddings, targets = downsample(
-            features, embeddings, targets, cfg.data.resolution
-        )
+            # preload the next batch, if we're not on the last batch
+            if i < epoch_len - 1:
+                dataloader.preload(epoch_indices[i + 1])
 
-        # Forward pass
-        outputs = model(features, embeddings)
+            # Cast precisions:
+            features, embeddings = cast_precisions(features, embeddings, precision)
 
-        metrics = metrics_fn(outputs, targets, dist_manager)
+            with context:
+                outputs = model(features, embeddings)
+                loss = loss_fn(outputs, targets, others, cfg.data.mode)
 
-        if i == 0:
-            total_metrics = metrics
-        else:
-            total_metrics = {
-                k: total_metrics[k] + metrics[k].item() for k in metrics.keys()
-            }
+            metrics = metrics_fn(outputs, targets, others, dist_manager, cfg.data.mode)
 
-        loss = loss_fn(outputs, targets)
+            optimizer.zero_grad()
+            if precision == "float16" and scaler is not None:
+                scaler.scale(loss).backward()
+                scaler.step(optimizer)
+                scaler.update()
+            else:
+                loss.backward()
+                optimizer.step()
 
-        # Backward pass
-        optimizer.zero_grad()
-        loss.backward()
-        optimizer.step()
-        scheduler.step()
+            if not isinstance(scheduler, torch.optim.lr_scheduler.ReduceLROnPlateau):
+                scheduler.step()
 
-        end_time = time.time()
-        duration = end_time - start_time
-        start_time = end_time
+            end_time = time.time()
 
-        images_per_second = 1 / duration
+            # Logging
+            total_loss += loss.item()
 
-        # Logging
-        total_loss += loss.item()
+            if i == 0:
+                total_metrics = metrics
+            else:
+                total_metrics = {
+                    k: total_metrics[k] + metrics[k].item() for k in metrics.keys()
+                }
 
-        logger.info(
-            f"Epoch {epoch} [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
-        )
-        if dist_manager.rank == 0:
-            writer.add_scalar(
-                "batch/learning_rate",
-                optimizer.param_groups[0]["lr"],
-                i + epoch_len * epoch,
-            )
-            writer.add_scalar("batch/loss", loss.item(), i + epoch_len * epoch)
-            writer.add_scalar(
-                "batch/throughpu_per_gpu", images_per_second, i + epoch_len * epoch
+            duration = end_time - start_time
+            start_time = end_time
+            images_per_second = 1 / duration
+
+            logger.info(
+                f"Epoch {epoch} [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
             )
-            for metric_name, metric_value in metrics.items():
+            if dist_manager.rank == 0:
+                writer.add_scalar(
+                    "batch/learning_rate",
+                    optimizer.param_groups[0]["lr"],
+                    i + epoch_len * epoch,
+                )
+                writer.add_scalar("batch/loss", loss.item(), i + epoch_len * epoch)
                 writer.add_scalar(
-                    f"batch/{metric_name}", metric_value, i + epoch_len * epoch
+                    "batch/throughpu_per_gpu", images_per_second, i + epoch_len * epoch
                 )
+                for metric_name, metric_value in metrics.items():
+                    writer.add_scalar(
+                        f"batch/{metric_name}", metric_value, i + epoch_len * epoch
+                    )
 
     avg_loss = total_loss / epoch_len
     avg_metrics = {k: v / epoch_len for k, v in total_metrics.items()}
@@ -192,7 +334,16 @@ def train_epoch(
 
 
 @profile
-def val_epoch(dataloader, sampler, model, logger, val_writer, epoch, cfg, dist_manager):
+def val_epoch(
+    dataloader,
+    sampler,
+    model,
+    logger,
+    val_writer,
+    epoch,
+    cfg,
+    dist_manager,
+):
     """Validation for one epoch
 
     Args:
@@ -212,22 +363,32 @@ def val_epoch(dataloader, sampler, model, logger, val_writer, epoch, cfg, dist_m
 
     epoch_indices = list(sampler)
     epoch_len = len(epoch_indices)
+    precision = getattr(cfg.training, "precision", "float32")
+    context = get_autocast_context(precision)
 
     start_time = time.time()
     with torch.no_grad():  # Disable gradient computation
         for i, batch_idx in enumerate(epoch_indices):
             batch = dataloader[batch_idx]
             # Get data from batch
-            features, embeddings, targets = preprocess_data(batch)
-
-            features, embeddings, targets = downsample(
-                features, embeddings, targets, cfg.data.resolution
-            )
+            if cfg.data.mode == "surface":
+                features, embeddings, targets, others = preprocess_surface_data(batch)
+                features, embeddings, targets = downsample_surface(
+                    features, embeddings, targets, cfg.data.resolution
+                )
+            elif cfg.data.mode == "volume":
+                features, embeddings, targets, others = preprocess_volume_data(batch)
+                features, embeddings, targets = downsample_volume(
+                    features, embeddings, targets, cfg.data.resolution
+                )
+            else:
+                raise ValueError(f"Unknown data mode: {cfg.data.mode}")
 
-            # Forward pass
-            outputs = model(features, embeddings)
+            with context:
+                outputs = model(features, embeddings)
+                loss = loss_fn(outputs, targets, others, cfg.data.mode)
 
-            metrics = metrics_fn(outputs, targets, dist_manager)
+            metrics = metrics_fn(outputs, targets, others, dist_manager, cfg.data.mode)
 
             if i == 0:
                 total_metrics = metrics
@@ -236,15 +397,13 @@ def val_epoch(dataloader, sampler, model, logger, val_writer, epoch, cfg, dist_m
                     k: total_metrics[k] + metrics[k].item() for k in metrics.keys()
                 }
 
-            loss = loss_fn(outputs, targets)
+            # Logging
+            total_loss += loss.item()
 
             end_time = time.time()
             duration = end_time - start_time
             start_time = end_time
 
-            # Logging
-            total_loss += loss.item()
-
             logger.info(
                 f"Val [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
             )
@@ -301,6 +460,7 @@ def main(cfg: DictConfig):
 
     # Set up model
     model = hydra.utils.instantiate(cfg.model)
+
     model.to(dist_manager.device)
 
     if dist_manager.world_size > 1:
@@ -324,8 +484,8 @@ def main(cfg: DictConfig):
         placements=placements,
         max_workers=cfg.data.max_workers,
         pin_memory=cfg.data.pin_memory,
-        keys_to_read=cfg.data.surface_keys,
-        large_keys=None,
+        keys_to_read=cfg.data.data_keys,
+        large_keys=cfg.data.large_keys,
     )
 
     # Validation dataset
@@ -335,8 +495,8 @@ def main(cfg: DictConfig):
         placements=placements,
         max_workers=cfg.data.max_workers,
         pin_memory=cfg.data.pin_memory,
-        keys_to_read=cfg.data.surface_keys,
-        large_keys=None,
+        keys_to_read=cfg.data.data_keys,
+        large_keys=cfg.data.large_keys,
     )
 
     # Set up distributed samplers
@@ -358,15 +518,29 @@ def main(cfg: DictConfig):
 
     # Set up optimizer and scheduler
     optimizer = hydra.utils.instantiate(cfg.optimizer, params=model.parameters())
-    # Set up OneCycleLR learning rate scheduler
-    scheduler = torch.optim.lr_scheduler.OneCycleLR(
-        optimizer,
-        max_lr=cfg.optimizer.lr,
-        total_steps=len(list(train_sampler)) * cfg.training.num_epochs,
-        pct_start=0.3,
-        div_factor=25.0,
-        final_div_factor=1e4,
-    )
+
+    # Set up learning rate scheduler based on config
+    scheduler_cfg = cfg.scheduler
+    scheduler_name = scheduler_cfg.name
+    scheduler_params = dict(scheduler_cfg.params)
+
+    if scheduler_name == "OneCycleLR":
+        scheduler_params.setdefault("max_lr", cfg.optimizer.lr)
+        # Dynamically compute total_steps
+        total_steps = len(list(train_sampler)) * cfg.training.num_epochs
+        scheduler_params["total_steps"] = total_steps
+        scheduler = torch.optim.lr_scheduler.OneCycleLR(optimizer, **scheduler_params)
+    elif scheduler_name == "ReduceLROnPlateau":
+        scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(
+            optimizer, **scheduler_params
+        )
+    elif scheduler_name == "StepLR":
+        scheduler = torch.optim.lr_scheduler.StepLR(optimizer, **scheduler_params)
+    else:
+        raise ValueError(f"Unknown scheduler: {scheduler_name}")
+
+    precision = getattr(cfg.training, "precision", "float32")
+    scaler = GradScaler() if precision == "float16" else None
 
     ckpt_args = {
         "path": f"{cfg.output_dir}/runs/{cfg.run_id}/checkpoints",
@@ -376,6 +550,9 @@ def main(cfg: DictConfig):
     }
     loaded_epoch = load_checkpoint(device=dist_manager.device, **ckpt_args)
 
+    if cfg.training.compile:
+        model = torch.compile(model)
+
     # Training loop
     logger.info("Starting training...")
     for epoch in range(loaded_epoch, cfg.training.num_epochs):
@@ -383,6 +560,7 @@ def main(cfg: DictConfig):
         train_sampler.set_epoch(epoch)
         val_sampler.set_epoch(epoch)
 
+        start_time = time.time()
         # Training phase
         train_loss = train_epoch(
             train_dataset,
@@ -395,8 +573,12 @@ def main(cfg: DictConfig):
             epoch,
             cfg,
             dist_manager,
+            scaler,
         )
+        end_time = time.time()
+        train_duration = end_time - start_time
 
+        start_time = time.time()
         # Validation phase
         val_loss = val_epoch(
             val_dataset,
@@ -408,16 +590,23 @@ def main(cfg: DictConfig):
             cfg,
             dist_manager,
         )
+        end_time = time.time()
+        val_duration = end_time - start_time
 
         # Log epoch results
         logger.info(
-            f"Epoch [{epoch}/{cfg.training.num_epochs}] Train Loss: {train_loss:.6f} Val Loss: {val_loss:.6f}"
+            f"Epoch [{epoch}/{cfg.training.num_epochs}] Train Loss: {train_loss:.6f} [duration: {train_duration:.2f}s] Val Loss: {val_loss:.6f} [duration: {val_duration:.2f}s]"
         )
 
         # save checkpoint
         if epoch % cfg.training.save_interval == 0 and dist_manager.rank == 0:
             save_checkpoint(**ckpt_args, epoch=epoch)
 
+        if scheduler_name == "ReduceLROnPlateau":
+            scheduler.step(val_loss)
+        else:
+            scheduler.step()
+
     logger.info("Training completed!")
 
 
@@ -428,11 +617,12 @@ def launch(cfg: DictConfig):
     Args:
         cfg: Hydra configuration object
     """
-    # profiler = Profiler()
+    profiler = Profiler()
+    # profiler.enable("torch")
     # profiler.enable("line_profiler")
-    # profiler.initialize()
+    profiler.initialize()
     main(cfg)
-    # profiler.finalize()
+    profiler.finalize()
 
 
 if __name__ == "__main__":

From 8c95135fe9619a00cca867a37e7fa456ae9083d9 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Fri, 18 Jul 2025 13:35:27 +0000
Subject: [PATCH 10/23] Updating transolver example further, and including more
 readme info

---
 .../transolver/README.md                      | 232 ++++++++++--------
 .../transolver/conf/train_surface.yaml        |  28 ++-
 .../transolver/conf/train_volume.yaml         |  14 +-
 .../transolver/datapipe.py                    |  65 ++++-
 .../external_aerodynamics/transolver/train.py |  52 ++--
 5 files changed, 230 insertions(+), 161 deletions(-)

diff --git a/examples/cfd/external_aerodynamics/transolver/README.md b/examples/cfd/external_aerodynamics/transolver/README.md
index 363cc5f037..d366f34843 100644
--- a/examples/cfd/external_aerodynamics/transolver/README.md
+++ b/examples/cfd/external_aerodynamics/transolver/README.md
@@ -1,130 +1,162 @@
 # Transolver CFD Example: Code Overview
 
-This directory contains the core components for training and evaluating a Transolver
-model for external aerodynamics CFD problems.  The Transolver model is an adaptation
-of the Attention mechanism to encourage models to learn meaningful representations.
-In each PhysicsAttention layer, the input points are projected onto state vectors via
-learnable transformations and weights.  These learnable transformations are in turn
-used to calculate self-attention between all state vectors, and the weights are reused
-to project states back to each input point.
-
-Through a series of PhysicsAttention layers, the Transolver model learns high quality
-projections from functional input space to output space.  The PhysicsNeMo implementation
-of Transolver is an equivalent implementation of the original model architecture
-([https://github.com/thuml/Transolver](https://github.com/thuml/Transolver)), with
-modifications to improve numerical stability and support NVIDIA TransformerEngine.
-
-The Transolver training recipe reuses the same input datasets as DoMINO - please see
-the external_aerodynamics example for DoMINO for more information, as well as
-[PhysicsNeMo Curator](https://github.com/NVIDIA/physicsnemo-curator) for information
-on producing the data if needed.  The Transolver model does not require the neighbor
-points as input to the model, nor does it require any structure to the input points
-(graph connections, edges, etc.).  Consequently, the datapipe is significantly simpler
-than for DoMINO - the entire datapipe is encapulated here in the examples.
-
-Below is a high-level overview of the main files and their roles in the workflow.
+This directory contains the essential components for training and evaluating a
+Transolver model tailored to external aerodynamics CFD problems. The Transolver model
+adapts the Attention mechanism, encouraging the learning of meaningful representations.
+In each PhysicsAttention layer, input points are projected onto state vectors through
+learnable transformations and weights. These transformations are then used to compute
+self-attention among all state vectors, and the same weights are reused to project
+states back to each input point.
+
+By stacking multiple PhysicsAttention layers, the Transolver model learns to map from
+the functional input space to the output space with high fidelity. The PhysicsNeMo
+implementation closely follows the original Transolver architecture
+([https://github.com/thuml/Transolver](https://github.com/thuml/Transolver)), but
+introduces modifications for improved numerical stability and compatibility with NVIDIA
+TransformerEngine.
+
+The training workflow for Transolver leverages the same input datasets as DoMINO. For
+more information on the datasets, refer to the external_aerodynamics example for DoMINO
+and the [PhysicsNeMo Curator](https://github.com/NVIDIA/physicsnemo-curator) for data
+preparation guidance. Unlike DoMINO, Transolver does not require neighbor points or any
+explicit structure (such as graph connections or edges) in the input data, resulting in
+a much simpler datapipe that is fully encapsulated within these examples.
+
+Below, we provide a high-level overview of the main files and their roles in the
+workflow.
 
 ---
 
-## 1. `conf/train.yaml`
-
-**Purpose:**  
-Configuration file for training runs.  
-**Contents:**  
-
-- **Output and run settings:** Output directory, run ID, random seed, and precision.
-- **Training parameters:** Number of epochs, checkpoint intervals, and whether to use
-  compilation.
-- **Model configuration:** Architecture details (input/output dimensions, number of
-  layers, embedding size, attention heads, activation, etc.).
-- **Optimizer settings:** Type (AdamW), learning rate, weight decay, and optimizer
-  hyperparameters.
-- **Data settings:** Paths to training/validation datasets, worker/thread settings,
-  memory pinning, and which data keys to load.
-- **Logging:** Logging level and format.
+## 1. `src/train.py`
+
+This script serves as the main entry point for training the DoMINO/Transolver model,
+utilizing distributed data parallelism. The training loop processes the entire dataset,
+computing a simple point-wise relative L2 loss for either surface or volume data, and
+includes downsampling to handle the high native mesh resolutions.
+
+The script is designed for multi-GPU training using PyTorch’s DistributedDataParallel,
+and it loads datasets through a custom datapipe that supports distributed sampling.
+Model instantiation is flexible, supporting either surface or volume predictions
+as specified in the configuration.
+
+The script supports mixed-precision training with gradient
+scaling, and it manages checkpointing with the physicsnemo checkpointing utils.
+Throughout training, metrics are logged to both TensorBoard and the console and
+validation is performed after each epoch.
+
+The script can be launched on a single GPU with
+
+```bash
+python train.py --config-name train_surface
+```
+
+or, for multi-GPU training, use `torchrun` or other distributed job launch tools.
+
+Example output for one epoch of the script, in an 8 GPU run, looks like:
+
+```default
+[2025-07-17 14:27:36,040][training][INFO] - Epoch 47 [0/54] Loss: 0.117565 Duration: 0.78s
+[2025-07-17 14:27:36,548][training][INFO] - Epoch 47 [1/54] Loss: 0.109625 Duration: 0.51s
+[2025-07-17 14:27:37,048][training][INFO] - Epoch 47 [2/54] Loss: 0.122574 Duration: 0.50s
+[2025-07-17 14:27:37,556][training][INFO] - Epoch 47 [3/54] Loss: 0.125667 Duration: 0.51s
+[2025-07-17 14:27:38,063][training][INFO] - Epoch 47 [4/54] Loss: 0.101863 Duration: 0.51s
+[2025-07-17 14:27:38,547][training][INFO] - Epoch 47 [5/54] Loss: 0.113324 Duration: 0.48s
+[2025-07-17 14:27:39,054][training][INFO] - Epoch 47 [6/54] Loss: 0.115478 Duration: 0.51s
+...[remove for brevity]...
+[2025-07-17 14:28:00,662][training][INFO] - Epoch 47 [49/54] Loss: 0.107935 Duration: 0.49s
+[2025-07-17 14:28:01,178][training][INFO] - Epoch 47 [50/54] Loss: 0.100087 Duration: 0.52s
+[2025-07-17 14:28:01,723][training][INFO] - Epoch 47 [51/54] Loss: 0.097733 Duration: 0.55s
+[2025-07-17 14:28:02,194][training][INFO] - Epoch 47 [52/54] Loss: 0.116489 Duration: 0.47s
+[2025-07-17 14:28:02,605][training][INFO] - Epoch 47 [53/54] Loss: 0.104865 Duration: 0.41s
+
+Epoch 47 Average Metrics:
++-------------+---------------------+
+|   Metric    |    Average Value    |
++-------------+---------------------+
+| l2_pressure | 0.20262257754802704 |
+| l2_shear_x  | 0.2623567283153534  |
+| l2_shear_y  | 0.35603201389312744 |
+| l2_shear_z  | 0.38965049386024475 |
++-------------+---------------------+
+
+[2025-07-17 14:28:02,834][training][INFO] - Val [0/6] Loss: 0.114801 Duration: 0.22s
+[2025-07-17 14:28:03,074][training][INFO] - Val [1/6] Loss: 0.111632 Duration: 0.24s
+[2025-07-17 14:28:03,309][training][INFO] - Val [2/6] Loss: 0.105342 Duration: 0.23s
+[2025-07-17 14:28:03,537][training][INFO] - Val [3/6] Loss: 0.111033 Duration: 0.23s
+[2025-07-17 14:28:03,735][training][INFO] - Val [4/6] Loss: 0.099963 Duration: 0.20s
+[2025-07-17 14:28:03,903][training][INFO] - Val [5/6] Loss: 0.092340 Duration: 0.17s
+
+Epoch 47 Validation Average Metrics:
++-------------+---------------------+
+|   Metric    |    Average Value    |
++-------------+---------------------+
+| l2_pressure | 0.19346082210540771 |
+| l2_shear_x  | 0.26041051745414734 |
+| l2_shear_y  | 0.3589216470718384  |
+| l2_shear_z  |  0.370105117559433  |
++-------------+---------------------+
+```
 
 ---
 
-## 2. `src/train.py`
-
-**Purpose:**  
-Main training script for the DoMINO/Transolver model using distributed data parallelism.
+## 2. `conf/train_volume.yaml` and `conf/train_surface.yaml`
 
-**Key Features:**
-
-- **Distributed Training:** Uses PyTorch DistributedDataParallel for multi-GPU training.
-- **Data Loading:** Loads datasets using custom datapipe, supports distributed sampling.
-- **Model Instantiation:** Builds the model based on config, supports both surface and
-  volume predictions.
-- **Loss Calculation:** Computes losses for volume, surface, and integral quantities
-  (lift/drag).
-- **Mixed Precision:** Supports mixed-precision training with gradient scaling.
-- **Checkpointing:** Automatically loads/saves checkpoints and tracks best validation
-  loss.
-- **Logging:** Logs metrics to TensorBoard and console, including GPU memory usage.
-- **Validation:** Evaluates model on validation set after each epoch.
+These configuration files define all the settings required for a training run. They
+specify output directories, run identifiers, random seeds, and precision settings, as
+well as the number of epochs, checkpoint intervals, and whether to use compilation. The
+model architecture is described here, including input and output dimensions, the number
+of layers, embedding size, attention heads, and activation functions. Optimizer
+settings such as the type (AdamW), learning rate, weight decay, and other
+hyperparameters are also included. Data-related settings cover paths to the training and
+validation datasets, worker and thread counts, memory pinning, and which data keys to
+load. Finally, logging preferences are set, controlling the level and format of output.
 
 ---
 
 ## 3. `loss.py`
 
-**Purpose:**  
-Defines loss functions for training the Transolver model.
-
-**Key Features:**
-
-- **Surface Loss:** Computes MSE or RMSE for both pressure (scalar) and wall shear
-  (vector) components, handling them separately and combining the results.
-- **Integral Losses:** Implements physics-based losses for lift and drag, using surface
-  integrals of predicted and true values, weighted by area, normals, and stream
-  velocity.
-- **Modularity:** Loss functions are modular and can be combined or extended for
-  different training objectives.
+This file defines the loss functions used during Transolver training, primarily
+focusing on a relative L2 loss. For surface data, it computes mean squared error (MSE)
+or root mean squared error (RMSE) for both pressure (a scalar) and wall shear (a
+vector), handling each component separately before combining the results. The file also
+implements physics-based losses for lift and drag, using surface integrals of predicted
+and true values, weighted appropriately by area, normals, and stream velocity. The loss
+functions are designed to be modular, allowing for easy combination or extension to
+support different training objectives.
 
 ---
 
 ## 4. `metrics.py`
 
-**Purpose:**  
-Defines evaluation metrics for model predictions.
-
-**Key Features:**
-
-- **Distributed Reduction:** Aggregates metrics across distributed processes.
-- **Surface Metrics:** Computes normalized L2 errors for pressure and shear components,
-  after unnormalizing predictions and targets.
-- **Extensibility:** Can be extended for additional metrics or domains.
+Evaluation metrics for model predictions are defined here. The metrics are designed to
+aggregate results across distributed processes, ensuring consistency in multi-GPU
+setups. For surface predictions, the script computes normalized L2 errors for both
+pressure and shear components, after unnormalizing the predictions and targets. The
+structure of the code allows for straightforward extension to additional metrics or
+application domains as needed.
 
 ---
 
 ## 5. `datapipe.py`
 
-**Purpose:**  
-Implements a PyTorch dataset for efficient loading of large CFD datasets stored in Zarr
-format.
-
-**Key Features:**
-
-- **Chunk-Aligned I/O:** Reads large arrays in chunk-aligned fashion for efficiency,
-  using threads for parallel I/O.
-- **Flexible Key Loading:** Allows specifying which data keys to load and which are
-  considered "large" (for chunked reading).
-- **Pinned Memory:** Optionally allocates pinned memory for faster GPU transfers.
-- **Prefetching:** Supports asynchronous preloading of samples to overlap I/O and
-  computation.
+Efficient loading of large CFD datasets stored in Zarr format is handled by this file,
+which implements a PyTorch dataset. The datapipe reads large arrays in a chunk-aligned
+manner for optimal performance, leveraging threads for parallel I/O. It offers
+flexibility in specifying which data keys to load and which are considered “large” (and
+thus read in chunks). For faster GPU transfers, it can allocate pinned memory, and it
+supports asynchronous prefetching of samples to overlap I/O with computation, further
+improving throughput.
 
 ---
 
 ## Summary
 
-- **`train.yaml`**: All configuration for model, data, optimizer, and logging.
-- **`train.py`**: Orchestrates distributed training, validation, checkpointing, and
-  logging.
-- **`loss.py`**: Physics-informed and standard loss functions for model training.
-- **`metrics.py`**: Evaluation metrics for model performance, with distributed support.
-- **`datapipe.py`**: High-performance, domain-parallel data loading from Zarr files for
-  large-scale CFD datasets.
-
----
+In summary, the configuration files (`train.yaml`) centralize all settings for the
+model, data, optimizer, and logging. The main training script (`train.py`) orchestrates
+distributed training, validation, checkpointing, and logging. Loss functions in
+`loss.py` combine physics-informed and standard approaches, while `metrics.py` provides
+robust evaluation metrics with distributed support. Finally, `datapipe.py` ensures
+high-performance, domain-parallel data loading from Zarr files, enabling large-scale
+CFD training.
 
-For more details, refer to the docstrings and comments within each file.
+For further details, consult the docstrings and comments within each file.
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
index db56902f48..3c01ef57fe 100644
--- a/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
@@ -17,7 +17,7 @@
 
 
 output_dir: "runs"
-run_id: "surface/ReduceLR/No-TE/float16/"
+run_id: "surface/ReduceLR/TE/float16/"
 
 # Training configuration
 training:
@@ -44,24 +44,26 @@ model:
   unified_pos: false # Whether to use unified positional embeddings
   ref: 8            # Reference dimension for unified pos
   structured_shape: null
-  use_te: false      # Use transformer engine
+  use_te: true      # Use transformer engine
   Time_Input: false # Whether to use time embeddings
 
-scheduler:
-  name: "OneCycleLR"
-  params:
-    final_div_factor: 1e4
-
 # scheduler:
-#   name: "ReduceLROnPlateau"
+#   name: "OneCycleLR"
 #   params:
-#     min_lr: 1e-6
+#     final_div_factor: 1e4
+
+# StepLR scheduler: Decays the learning rate by gamma every step_size epochs
+scheduler:
+  name: "StepLR"
+  params:
+    step_size: 100    # Decay every 100 epochs (set X as desired)
+    gamma: 0.5        # Decay factor
 
 
 # Optimizer configuration
 optimizer:
   _target_: torch.optim.AdamW
-  lr: 3.0e-4
+  lr: 1.0e-3
   weight_decay: 1.0e-4
   betas: [0.9, 0.999]
   eps: 1.0e-8
@@ -69,12 +71,12 @@ optimizer:
 # Data configuration
 data:
   train:
-    data_path: /user_data/datasets/domino/train/
+    data_path: /group_data/datasets/drivaer_aws/domino/train/
   val:
-    data_path: /user_data/datasets/domino/val/
+    data_path: /group_data/datasets/drivaer_aws/domino/val/
   max_workers: 64
   pin_memory: true
-  # resolution: 262_144
+  # resolution: 10000
   resolution: 325_000
   mode: surface
   data_keys:
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
index 889a57c0ff..f3bd7275d2 100644
--- a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
@@ -17,12 +17,12 @@
 
 
 output_dir: "runs"
-run_id: "volume/OneCycleLR/TE/float32/"
+run_id: "volume/OneCycleLR/noTE/float32/"
 
 # Training configuration
 training:
   precision: float32 # float32, float16, bfloat16
-  num_epochs: 200
+  num_epochs: 1
   save_interval: 25  # Save checkpoint every N epochs
   seed: 42
   output_dir: "runs"
@@ -44,7 +44,7 @@ model:
   unified_pos: false # Whether to use unified positional embeddings
   ref: 8            # Reference dimension for unified pos
   structured_shape: null
-  use_te: true      # Use transformer engine
+  use_te: false      # Use transformer engine
   Time_Input: false # Whether to use time embeddings
 
 scheduler:
@@ -69,10 +69,10 @@ optimizer:
 # Data configuration
 data:
   train:
-    data_path: /user_data/datasets/domino/train/
+    data_path: /group_data/datasets/drivaer_aws/domino/train/
   val:
-    data_path: /user_data/datasets/domino/val/
-  max_workers: 64
+    data_path: /group_data/datasets/drivaer_aws/domino/val/
+  max_workers: 8
   pin_memory: true
   # resolution: 262_144
   resolution: 200_000
@@ -82,7 +82,7 @@ data:
     - "volume_fields"
     - "air_density"
     - "stream_velocity"
-  large_keys:
+  large_keys: 
     - "volume_fields"
     - "volume_mesh_centers"
 
diff --git a/examples/cfd/external_aerodynamics/transolver/datapipe.py b/examples/cfd/external_aerodynamics/transolver/datapipe.py
index 1c9e168c01..6d42b349f6 100644
--- a/examples/cfd/external_aerodynamics/transolver/datapipe.py
+++ b/examples/cfd/external_aerodynamics/transolver/datapipe.py
@@ -175,6 +175,9 @@ def __init__(
         self.keys_to_read = keys_to_read or set()
         self.large_keys = large_keys or set()
 
+        self.data_loader_stream = torch.cuda.Stream()
+        self.gpu_transfer_compete = None
+
         # Validate distributed configuration
         if self.device_mesh is not None:
             if self.device_mesh.ndim != 1:
@@ -458,10 +461,25 @@ def _move_to_gpu(self, data: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]
         """
         result = {}
 
-        for key, array in data.items():
-            # Move to GPU if available
-            if self.dm.cuda:
-                result[key] = data[key].to(self.dm.device, non_blocking=True)
+        # This is old code.
+        # Thinking about ways to reduce CPU overhead.
+        # If we could avoid the data reading step, it would be easier.
+        # total_size = 0
+        # for key, array in data.items():
+        #     size = array.numel() * array.element_size()
+        #     total_size += size
+        #     print(f"Size in memory of {key}: {size / 1024 / 1024 / 1024} GB")
+        # print(f"Total size in memory: {total_size / 1024 / 1024 / 1024} GB")
+
+        with torch.cuda.stream(self.data_loader_stream):
+
+            for key, array in data.items():
+                # Move to GPU if available
+                if self.dm.cuda:
+                    result[key] = data[key].to(self.dm.device, non_blocking=True)
+
+            self.gpu_transfer_compete = torch.cuda.Event()
+            self.gpu_transfer_compete.record(self.data_loader_stream)
 
         return result
 
@@ -526,6 +544,8 @@ def _preload_worker():
                 filename = self.filenames[idx]
                 filepath = self.data_path / filename
                 data = self._read_zarr_file(filepath)
+                # Convert to torch tensors
+                data = self._move_to_gpu(data)
                 self._preload_result = (idx, data)
             except Exception as e:
                 self._preload_exception = e
@@ -555,6 +575,25 @@ def get_preloaded(self) -> Tuple[int, Dict[str, np.ndarray]]:
 
         return self._preload_result
 
+    def __iter__(self):
+        self.i = 0
+        return self
+
+    def __next__(self):
+        """
+        This supports the DALI iterator.
+        """
+        if self.i >= len(self.filenames):
+            self.i = 0
+            raise StopIteration
+        data = self._read_zarr_file(self.data_path / self.filenames[self.i])
+
+        self.i += 1
+        return tuple(data[key].unsqueeze(0) for key in self.keys_to_read)
+
+    def __len__(self):
+        return len(self.filenames)
+
     @profile
     def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
         """
@@ -579,27 +618,27 @@ def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
             else:
                 # Preloaded data is for a different idx, ignore it
                 data = self._read_zarr_file(self.data_path / self.filenames[idx])
+                data = self._move_to_gpu(data)
         else:
             filename = self.filenames[idx]
             filepath = self.data_path / filename
 
             # Read data from zarr file
             data = self._read_zarr_file(filepath)
+            data = self._move_to_gpu(data)
 
-        # filename = self.filenames[idx]
-        # filepath = self.data_path / filename
-
-        # # Read data from zarr file
-        # data = self._read_zarr_file(filepath)
+        # This blocks until the preprocessing has transferred to GPU
+        if self.gpu_transfer_compete is not None:
+            torch.cuda.current_stream().wait_event(self.gpu_transfer_compete)
 
-        # Convert to torch tensors
-        tensors = self._move_to_gpu(data)
+        # Add a batch index:
+        data = {key: value.unsqueeze(0) for key, value in data.items()}
 
         # Convert to ShardTensors if using domain parallelism
         if self.device_mesh is not None:
-            tensors = self._convert_to_shard_tensors(tensors)
+            data = self._convert_to_shard_tensors(data)
 
-        return tensors
+        return data
 
 
 # TODO: Additional features to consider implementing:
diff --git a/examples/cfd/external_aerodynamics/transolver/train.py b/examples/cfd/external_aerodynamics/transolver/train.py
index 26f348f04d..4626171493 100644
--- a/examples/cfd/external_aerodynamics/transolver/train.py
+++ b/examples/cfd/external_aerodynamics/transolver/train.py
@@ -97,8 +97,8 @@ def preprocess_surface_data(batch):
     ).to(torch.float32)
 
     # Normalize the surface fields:
-    norm_mean = targets.mean(dim=0)
-    norm_std = targets.std(dim=0)
+    norm_mean = targets.mean(dim=1)
+    norm_std = targets.std(dim=1)
     targets = (targets - norm_mean) / norm_std
     # fourier_sin_features = [
     #     torch.sin(mesh_centers * (2 ** i) * torch.pi)
@@ -113,7 +113,7 @@ def preprocess_surface_data(batch):
     sizes = batch["stl_areas"]
     centers = batch["stl_centers"]
 
-    total_weighted_position = torch.einsum("i,ij->j", sizes, centers)
+    total_weighted_position = torch.einsum("ki,kij->kj", sizes, centers)
     total_size = torch.sum(sizes)
     center_of_mass = total_weighted_position[None, ...] / total_size
 
@@ -129,8 +129,7 @@ def preprocess_surface_data(batch):
         ],
         dim=-1,
     )
-
-    node_features = node_features.unsqueeze(0).broadcast_to(embeddings.shape[0], -1)
+    node_features = node_features.unsqueeze(1).broadcast_to(1, embeddings.shape[1], -1)
 
     others = {
         "surface_areas": sizes,
@@ -171,20 +170,20 @@ def preprocess_volume_data(batch):
 @profile
 def downsample_surface(features, embeddings, targets, num_keep=1024):
     # Determine the number of samples to keep (e.g., 50% of original size)
-    num_samples = features.shape[0]
+    num_samples = features.shape[1]
     # Generate random indices to keep (faster for large num_samples)
     indices = torch.multinomial(
         torch.ones(num_samples, device=features.device), num_keep, replacement=False
     )
 
     # Use the same indices to downsample all tensors
-    downsampled_features = features[indices]
-    downsampled_embeddings = embeddings[indices]
-    downsampled_targets = targets[indices]
+    downsampled_features = features[:, indices]
+    downsampled_embeddings = embeddings[:, indices]
+    downsampled_targets = targets[:, indices]
 
-    downsampled_features = downsampled_features.unsqueeze(0)
-    downsampled_embeddings = downsampled_embeddings.unsqueeze(0)
-    downsampled_targets = downsampled_targets.unsqueeze(0)
+    # downsampled_features = downsampled_features.unsqueeze(0)
+    # downsampled_embeddings = downsampled_embeddings.unsqueeze(0)
+    # downsampled_targets = downsampled_targets.unsqueeze(0)
 
     return downsampled_features, downsampled_embeddings, downsampled_targets
 
@@ -192,19 +191,15 @@ def downsample_surface(features, embeddings, targets, num_keep=1024):
 @profile
 def downsample_volume(features, embeddings, targets, num_keep=1024):
     # Determine the number of samples to keep (e.g., 50% of original size)
-    num_samples = features.shape[0]
+    num_samples = features.shape[1]
     # The volume data is so large, that we'll sample randints
     # which will very rarely duplicate
     indices = torch.randint(0, num_samples, (num_keep,), device=features.device)
 
     # Use the same indices to downsample all tensors
-    downsampled_features = features[indices]
-    downsampled_embeddings = embeddings[indices]
-    downsampled_targets = targets[indices]
-
-    downsampled_features = downsampled_features.unsqueeze(0)
-    downsampled_embeddings = downsampled_embeddings.unsqueeze(0)
-    downsampled_targets = downsampled_targets.unsqueeze(0)
+    downsampled_features = features[:, indices]
+    downsampled_embeddings = embeddings[:, indices]
+    downsampled_targets = targets[:, indices]
 
     return downsampled_features, downsampled_embeddings, downsampled_targets
 
@@ -236,13 +231,14 @@ def train_epoch(
     total_loss = 0
     total_metrics = {}
 
-    epoch_indices = list(sampler)
+    epoch_indices = list(sampler) if sampler is not None else range(len(dataloader))
     epoch_len = len(epoch_indices)
     precision = getattr(cfg.training, "precision", "float32")
     context = get_autocast_context(precision)
     start_time = time.time()
     with Profiler():
         for i, batch_idx in enumerate(epoch_indices):
+
             batch = dataloader[batch_idx]
             # Get data from batch
             if cfg.data.mode == "surface":
@@ -259,7 +255,7 @@ def train_epoch(
                 raise ValueError(f"Unknown data mode: {cfg.data.mode}")
 
             # preload the next batch, if we're not on the last batch
-            if i < epoch_len - 1:
+            if i < epoch_len - 1 and sampler is not None:
                 dataloader.preload(epoch_indices[i + 1])
 
             # Cast precisions:
@@ -280,7 +276,7 @@ def train_epoch(
                 loss.backward()
                 optimizer.step()
 
-            if not isinstance(scheduler, torch.optim.lr_scheduler.ReduceLROnPlateau):
+            if not isinstance(scheduler, torch.optim.lr_scheduler.StepLR):
                 scheduler.step()
 
             end_time = time.time()
@@ -361,7 +357,7 @@ def val_epoch(
     total_loss = 0
     total_metrics = {}
 
-    epoch_indices = list(sampler)
+    epoch_indices = list(sampler) if sampler is not None else range(len(dataloader))
     epoch_len = len(epoch_indices)
     precision = getattr(cfg.training, "precision", "float32")
     context = get_autocast_context(precision)
@@ -478,6 +474,7 @@ def main(cfg: DictConfig):
     placements = None
 
     # Training dataset
+
     train_dataset = DomainParallelZarrDataset(
         data_path=cfg.data.train.data_path,
         device_mesh=device_mesh,
@@ -489,6 +486,7 @@ def main(cfg: DictConfig):
     )
 
     # Validation dataset
+
     val_dataset = DomainParallelZarrDataset(
         data_path=cfg.data.val.data_path,  # Assuming validation data path is configured
         device_mesh=device_mesh,
@@ -543,7 +541,7 @@ def main(cfg: DictConfig):
     scaler = GradScaler() if precision == "float16" else None
 
     ckpt_args = {
-        "path": f"{cfg.output_dir}/runs/{cfg.run_id}/checkpoints",
+        "path": f"{cfg.output_dir}/{cfg.run_id}/checkpoints",
         "optimizer": optimizer,
         "scheduler": scheduler,
         "models": model,
@@ -602,9 +600,7 @@ def main(cfg: DictConfig):
         if epoch % cfg.training.save_interval == 0 and dist_manager.rank == 0:
             save_checkpoint(**ckpt_args, epoch=epoch)
 
-        if scheduler_name == "ReduceLROnPlateau":
-            scheduler.step(val_loss)
-        else:
+        if scheduler_name == "StepLR":
             scheduler.step()
 
     logger.info("Training completed!")

From 858c5371a1aead05505e72b7c58fb55711c99adb Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Fri, 18 Jul 2025 13:12:37 -0700
Subject: [PATCH 11/23] Further integrate transformer engine into Transolver.

---
 .../models/transolver/Physics_Attention.py    | 19 +++++--
 physicsnemo/models/transolver/transolver.py   | 51 +++++++++++++------
 2 files changed, 50 insertions(+), 20 deletions(-)

diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index ab1409eb21..556ea28ec3 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -63,8 +63,13 @@ def __init__(self, dim, heads, dim_head, dropout, slice_num, use_te):
         self.softmax = nn.Softmax(dim=-1)
         self.dropout = nn.Dropout(dropout)
         self.temperature = nn.Parameter(torch.ones([1, heads, 1, 1]) * 0.5)
+        self.use_te = use_te
+
+        if self.use_te:
+            self.in_project_slice = te.Linear(dim_head, slice_num)
+        else:
+            self.in_project_slice = nn.Linear(dim_head, slice_num)
 
-        self.in_project_slice = nn.Linear(dim_head, slice_num)
         for l_i in [self.in_project_slice]:
             torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
         if not use_te:
@@ -73,7 +78,7 @@ def __init__(self, dim, heads, dim_head, dropout, slice_num, use_te):
             self.to_v = nn.Linear(dim_head, dim_head, bias=False)
         else:
             # These are used in the transformer engine pass function:
-            self.qkv_project = nn.Linear(dim_head, 3 * dim_head, bias=False)
+            self.qkv_project = te.Linear(dim_head, 3 * dim_head, bias=False)
             self.attn_fn = te.DotProductAttention(
                 num_attention_heads=self.heads,
                 kv_channels=self.dim_head,
@@ -82,9 +87,12 @@ def __init__(self, dim, heads, dim_head, dropout, slice_num, use_te):
                 softmax_scale=self.scale,
             )
 
-        self.use_te = use_te
+        if self.use_te:
+            self.out_linear = te.Linear(inner_dim, dim)
+        else:
+            self.out_linear = nn.Linear(inner_dim, dim)
 
-        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
+        self.out_dropout = nn.Dropout(dropout)
 
     def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
         """
@@ -216,7 +224,8 @@ def project_attention_outputs(
 
         out_x = torch.matmul(slice_weights, out_slice_token)
         out_x = rearrange(out_x, "b h n d -> b n (h d)")
-        return self.to_out(out_x)
+        out_x = self.out_linear(out_x)
+        return self.out_dropout(out_x)
 
     def forward(self, x: torch.Tensor) -> torch.Tensor:
         """
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index 6dbf16d3f7..51b953a2cb 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -72,7 +72,9 @@
 
 
 class MLP(nn.Module):
-    def __init__(self, n_input, n_hidden, n_output, n_layers=1, act="gelu", res=True):
+    def __init__(
+        self, n_input, n_hidden, n_output, n_layers=1, act="gelu", res=True, use_te=True
+    ):
         super(MLP, self).__init__()
 
         if act in ACTIVATION.keys():
@@ -84,22 +86,27 @@ def __init__(self, n_input, n_hidden, n_output, n_layers=1, act="gelu", res=True
         self.n_output = n_output
         self.n_layers = n_layers
         self.res = res
-        self.linear_pre = nn.Sequential(nn.Linear(n_input, n_hidden), act())
-        self.linear_post = nn.Linear(n_hidden, n_output)
-        self.linears = nn.ModuleList(
-            [
-                nn.Sequential(nn.Linear(n_hidden, n_hidden), act())
-                for _ in range(n_layers)
-            ]
-        )
+
+        self.act = act()
+
+        linear_layer = nn.Linear if not use_te else te.Linear
+
+        self.linear_pre = linear_layer(n_input, n_hidden)
+        self.linear_post = linear_layer(n_hidden, n_output)
+        # self.linears = nn.ModuleList(
+        #     [
+        #         nn.Sequential(linear_layer(n_hidden, n_hidden), act())
+        #         for _ in range(n_layers)
+        #     ]
+        # )
 
     def forward(self, x):
-        x = self.linear_pre(x)
-        for i in range(self.n_layers):
-            if self.res:
-                x = self.linears[i](x) + x
-            else:
-                x = self.linears[i](x)
+        x = self.act(self.linear_pre(x))
+        # for i in range(self.n_layers):
+        #     if self.res:
+        #         x = self.linears[i](x) + x
+        #     else:
+        #         x = self.linears[i](x)
         x = self.linear_post(x)
         return x
 
@@ -173,6 +180,18 @@ def __init__(
                 hidden_size=hidden_dim,
                 ffn_hidden_size=hidden_dim * mlp_ratio,
             )
+            # self.ln_mlp1 = nn.Sequential(
+            #     te.LayerNorm(hidden_dim),
+            #     MLP(
+            #         hidden_dim,
+            #         hidden_dim * mlp_ratio,
+            #         hidden_dim,
+            #         n_layers=0,
+            #         res=False,
+            #         act=act,
+            #         use_te=True
+            #     ),
+            # )
         else:
             self.ln_mlp1 = nn.Sequential(
                 nn.LayerNorm(hidden_dim),
@@ -183,6 +202,7 @@ def __init__(
                     n_layers=0,
                     res=False,
                     act=act,
+                    use_te=False,
                 ),
             )
         if self.last_layer:
@@ -377,6 +397,7 @@ def __init__(
             n_layers=0,
             res=False,
             act=act,
+            use_te=False,
         )
 
         self.Time_Input = Time_Input

From 221cab11768e6fa1db9d283c8f966bd65217f85f Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Fri, 18 Jul 2025 13:56:45 -0700
Subject: [PATCH 12/23] Update readme to point out matlab to npz conversion of
 fixed dataset.

---
 examples/cfd/darcy_transolver/README.md | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/examples/cfd/darcy_transolver/README.md b/examples/cfd/darcy_transolver/README.md
index 571f19bdc3..fcfa96da19 100644
--- a/examples/cfd/darcy_transolver/README.md
+++ b/examples/cfd/darcy_transolver/README.md
@@ -44,6 +44,10 @@ and the data path should be added when `Darcy_2D_fix` dataset is constructed.
 You can download the data
 [here](https://huggingface.co/datasets/lkuang/example_data).
 
+The `fix` dataset training (which uses a fixed dataset) requires you to convert
+data from matlab to numpy format, for faster startup of the training.  Just
+use the `convert_mat_to_npz.py` script to port your data.
+
 ## Model overview and architecture
 
 ## Getting Started

From 5ea778909b7b45b8534cd21871f1c82f6a31b6ce Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Fri, 18 Jul 2025 14:51:20 -0700
Subject: [PATCH 13/23] Update README and improve scripts

---
 .../transolver/README.md                      |  38 +--
 .../transolver/conf/train_surface.yaml        |  12 +-
 .../transolver/conf/train_volume.yaml         |  15 +-
 .../transolver/metrics.py                     | 114 ++++++-
 .../transolver/requirements.txt               |   6 +
 .../external_aerodynamics/transolver/train.py | 292 ++++++++++++------
 6 files changed, 323 insertions(+), 154 deletions(-)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/requirements.txt

diff --git a/examples/cfd/external_aerodynamics/transolver/README.md b/examples/cfd/external_aerodynamics/transolver/README.md
index d366f34843..80b6a00474 100644
--- a/examples/cfd/external_aerodynamics/transolver/README.md
+++ b/examples/cfd/external_aerodynamics/transolver/README.md
@@ -111,6 +111,9 @@ hyperparameters are also included. Data-related settings cover paths to the trai
 validation datasets, worker and thread counts, memory pinning, and which data keys to
 load. Finally, logging preferences are set, controlling the level and format of output.
 
+Note that, to use TransformerEngine, you must enable it directly in the model settings.
+TransformerEngine is incompatible with `torch.compile`.  
+
 ---
 
 ## 3. `loss.py`
@@ -118,11 +121,7 @@ load. Finally, logging preferences are set, controlling the level and format of
 This file defines the loss functions used during Transolver training, primarily
 focusing on a relative L2 loss. For surface data, it computes mean squared error (MSE)
 or root mean squared error (RMSE) for both pressure (a scalar) and wall shear (a
-vector), handling each component separately before combining the results. The file also
-implements physics-based losses for lift and drag, using surface integrals of predicted
-and true values, weighted appropriately by area, normals, and stream velocity. The loss
-functions are designed to be modular, allowing for easy combination or extension to
-support different training objectives.
+vector), handling each component separately before combining the results.
 
 ---
 
@@ -140,23 +139,16 @@ application domains as needed.
 ## 5. `datapipe.py`
 
 Efficient loading of large CFD datasets stored in Zarr format is handled by this file,
-which implements a PyTorch dataset. The datapipe reads large arrays in a chunk-aligned
+which implements like a PyTorch dataset. The datapipe reads large arrays in a chunk-aligned
 manner for optimal performance, leveraging threads for parallel I/O. It offers
 flexibility in specifying which data keys to load and which are considered “large” (and
-thus read in chunks). For faster GPU transfers, it can allocate pinned memory, and it
-supports asynchronous prefetching of samples to overlap I/O with computation, further
-improving throughput.
-
----
-
-## Summary
-
-In summary, the configuration files (`train.yaml`) centralize all settings for the
-model, data, optimizer, and logging. The main training script (`train.py`) orchestrates
-distributed training, validation, checkpointing, and logging. Loss functions in
-`loss.py` combine physics-informed and standard approaches, while `metrics.py` provides
-robust evaluation metrics with distributed support. Finally, `datapipe.py` ensures
-high-performance, domain-parallel data loading from Zarr files, enabling large-scale
-CFD training.
-
-For further details, consult the docstrings and comments within each file.
+thus read in chunks, vs. small arrays which are directly read in their entirety).
+For faster GPU transfers, it can allocate directly to pinned memory, then share buffers
+with numpy for streaming from disk directly to pinned memory.  The dataload has a
+prefetch utility, as well, enabling the dataloader to queue the next batch to GPU.
+
+CPU to GPU transfers are performed in a separate CUDA stream from the main computation,
+enabling async transfers overlapping with model training.  The main parameter to tune
+is the number of workers for the threading: too many, and you will introduce
+CPU overhead limiting the model performance.  Too few, and the dataload won't acheive
+peak throughput of dataloading.
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
index 3c01ef57fe..76b9c6e691 100644
--- a/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
@@ -17,7 +17,7 @@
 
 
 output_dir: "runs"
-run_id: "surface/ReduceLR/TE/float16/"
+run_id: "surface/ReduceLR/TE/float16-3/"
 
 # Training configuration
 training:
@@ -44,7 +44,7 @@ model:
   unified_pos: false # Whether to use unified positional embeddings
   ref: 8            # Reference dimension for unified pos
   structured_shape: null
-  use_te: true      # Use transformer engine
+  use_te: True      # Use transformer engine
   Time_Input: false # Whether to use time embeddings
 
 # scheduler:
@@ -56,14 +56,14 @@ model:
 scheduler:
   name: "StepLR"
   params:
-    step_size: 100    # Decay every 100 epochs (set X as desired)
+    step_size: 50     # Decay every 100 epochs (set X as desired)
     gamma: 0.5        # Decay factor
 
 
 # Optimizer configuration
 optimizer:
   _target_: torch.optim.AdamW
-  lr: 1.0e-3
+  lr: 3.0e-4
   weight_decay: 1.0e-4
   betas: [0.9, 0.999]
   eps: 1.0e-8
@@ -71,9 +71,9 @@ optimizer:
 # Data configuration
 data:
   train:
-    data_path: /group_data/datasets/drivaer_aws/domino/train/
+    data_path: /user_data/datasets/domino/train/
   val:
-    data_path: /group_data/datasets/drivaer_aws/domino/val/
+    data_path: /user_data/datasets/domino/val/
   max_workers: 64
   pin_memory: true
   # resolution: 10000
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
index f3bd7275d2..97f24b9134 100644
--- a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
@@ -17,16 +17,16 @@
 
 
 output_dir: "runs"
-run_id: "volume/OneCycleLR/noTE/float32/"
+run_id: "volume/OneCycleLR/noTE/bfloat16/"
 
 # Training configuration
 training:
-  precision: float32 # float32, float16, bfloat16
-  num_epochs: 1
+  precision: bfloat16 # float32, float16, bfloat16
+  num_epochs: 250
   save_interval: 25  # Save checkpoint every N epochs
   seed: 42
   output_dir: "runs"
-  compile: false
+  compile: true
 
 # Model configuration
 model:
@@ -69,12 +69,11 @@ optimizer:
 # Data configuration
 data:
   train:
-    data_path: /group_data/datasets/drivaer_aws/domino/train/
+    data_path: /user_data/datasets/domino/train/
   val:
-    data_path: /group_data/datasets/drivaer_aws/domino/val/
-  max_workers: 8
+    data_path: /user_data/datasets/domino/val/
+  max_workers: 32
   pin_memory: true
-  # resolution: 262_144
   resolution: 200_000
   mode: volume
   data_keys:
diff --git a/examples/cfd/external_aerodynamics/transolver/metrics.py b/examples/cfd/external_aerodynamics/transolver/metrics.py
index 2b3dfc1939..18e8ecfa2e 100644
--- a/examples/cfd/external_aerodynamics/transolver/metrics.py
+++ b/examples/cfd/external_aerodynamics/transolver/metrics.py
@@ -20,7 +20,19 @@
 from physicsnemo.distributed import DistributedManager
 
 
-def all_reduce_dict(metrics, dm):
+def all_reduce_dict(
+    metrics: dict[str, torch.Tensor], dm: DistributedManager
+) -> dict[str, torch.Tensor]:
+    """
+    Reduces a dictionary of metrics across all distributed processes.
+
+    Args:
+        metrics: Dictionary of metric names to torch.Tensor values.
+        dm: DistributedManager instance for distributed context.
+
+    Returns:
+        Dictionary of reduced metrics.
+    """
     # TODO - update this to use domains and not the full world
 
     if dm.world_size == 1:
@@ -34,16 +46,56 @@ def all_reduce_dict(metrics, dm):
     return metrics
 
 
-def metrics_fn(pred, target, others, dm, mode):
-    if mode == "surface":
-        return metrics_fn_surface(pred, target, others, dm)
-    elif mode == "volume":
-        return metrics_fn_volume(pred, target, others, dm)
-    else:
-        raise ValueError(f"Unknown data mode: {mode}")
+def metrics_fn(
+    pred: torch.Tensor,
+    target: torch.Tensor,
+    others: dict[str, torch.Tensor],
+    dm: DistributedManager,
+    mode: str,
+) -> dict[str, torch.Tensor]:
+    """
+    Computes metrics for either surface or volume data.
+
+    Args:
+        pred: Predicted values (unnormalized).
+        target: Target values (unnormalized).
+        others: Dictionary containing normalization statistics.
+        dm: DistributedManager instance for distributed context.
+        mode: Either "surface" or "volume".
+
+    Returns:
+        Dictionary of computed metrics.
+    """
+    with torch.no_grad():
+        if mode == "surface":
+            metrics = metrics_fn_surface(pred, target, others, dm)
+        elif mode == "volume":
+            metrics = metrics_fn_volume(pred, target, others, dm)
+        else:
+            raise ValueError(f"Unknown data mode: {mode}")
+
+        metrics = all_reduce_dict(metrics, dm)
+        return metrics
 
 
-def metrics_fn_volume(pred, target, others, dm):
+def metrics_fn_volume(
+    pred: torch.Tensor,
+    target: torch.Tensor,
+    others: dict[str, torch.Tensor],
+    dm: DistributedManager,
+) -> dict[str, torch.Tensor]:
+    """
+    Computes L2 volume metrics between prediction and target.
+
+    Args:
+        pred: Predicted values (normalized).
+        target: Target values (normalized).
+        others: Dictionary containing normalization statistics.
+        dm: DistributedManager instance for distributed context.
+
+    Returns:
+        Dictionary of L2 volume metrics.
+    """
     target = target * others["norm_std"] + others["norm_mean"]
     pred = pred * others["norm_std"] + others["norm_mean"]
 
@@ -58,18 +110,34 @@ def metrics_fn_volume(pred, target, others, dm):
     l2 = l2_num / l2_denom
 
     metrics = {
-        "l2_volume_1": torch.mean(l2[:, 0]),
-        "l2_volume_2": torch.mean(l2[:, 1]),
-        "l2_volume_3": torch.mean(l2[:, 2]),
-        "l2_volume_4": torch.mean(l2[:, 3]),
-        "l2_volume_5": torch.mean(l2[:, 4]),
+        "lr_volume_pressure": torch.mean(l2[:, 0]),
+        "l2_velocity_x": torch.mean(l2[:, 1]),
+        "l2_velocity_y": torch.mean(l2[:, 2]),
+        "l2_velocity_z": torch.mean(l2[:, 3]),
+        "l2_turb_visc": torch.mean(l2[:, 4]),
     }
 
     return metrics
 
 
-def metrics_fn_surface(pred, target, others, dm):
-
+def metrics_fn_surface(
+    pred: torch.Tensor,
+    target: torch.Tensor,
+    others: dict[str, torch.Tensor],
+    dm: DistributedManager,
+) -> dict[str, torch.Tensor]:
+    """
+    Computes L2 surface metrics between prediction and target.
+
+    Args:
+        pred: Predicted values (normalized).
+        target: Target values (normalized).
+        others: Dictionary containing normalization statistics.
+        dm: DistributedManager instance for distributed context.
+
+    Returns:
+        Dictionary of L2 surface metrics.
+    """
     # Unnormalize the surface values for L2:
     target = target * others["norm_std"] + others["norm_mean"]
     pred = pred * others["norm_std"] + others["norm_mean"]
@@ -94,5 +162,17 @@ def metrics_fn_surface(pred, target, others, dm):
     return metrics
 
 
-def metrics_fn_surface_pressure(pred, target):
+def metrics_fn_surface_pressure(
+    pred: torch.Tensor, target: torch.Tensor
+) -> torch.Tensor:
+    """
+    Computes mean squared error between predicted and target surface pressure.
+
+    Args:
+        pred: Predicted surface pressure.
+        target: Target surface pressure.
+
+    Returns:
+        Mean squared error as a torch.Tensor.
+    """
     return torch.mean((pred - target) ** 2.0)
diff --git a/examples/cfd/external_aerodynamics/transolver/requirements.txt b/examples/cfd/external_aerodynamics/transolver/requirements.txt
new file mode 100644
index 0000000000..e0414bda88
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/requirements.txt
@@ -0,0 +1,6 @@
+hydra-core
+tabulate
+tensorboard
+termcolor
+einops
+transformer_engine[pytorch]
\ No newline at end of file
diff --git a/examples/cfd/external_aerodynamics/transolver/train.py b/examples/cfd/external_aerodynamics/transolver/train.py
index 4626171493..954dd42a2d 100644
--- a/examples/cfd/external_aerodynamics/transolver/train.py
+++ b/examples/cfd/external_aerodynamics/transolver/train.py
@@ -37,8 +37,8 @@
 from contextlib import nullcontext
 from torch.amp import autocast, GradScaler
 
-# import transformer_engine.pytorch as te
-# from transformer_engine.common.recipe import Format, DelayedScaling
+import transformer_engine.pytorch as te
+from transformer_engine.common.recipe import Format, DelayedScaling
 
 
 def get_autocast_context(precision: str) -> nullcontext:
@@ -55,11 +55,12 @@ def get_autocast_context(precision: str) -> nullcontext:
         return autocast("cuda", dtype=torch.float16)
     elif precision == "bfloat16":
         return autocast("cuda", dtype=torch.bfloat16)
-    # elif precision == "float8":
-    #     print("Using float8 autocast")
-    #     fp8_format = Format.HYBRID
-    #     fp8_recipe = DelayedScaling(fp8_format=fp8_format, amax_history_len=16, amax_compute_algo="max")
-    #     return te.fp8_autocast(enabled=True, fp8_recipe=fp8_recipe)
+    elif precision == "float8":
+        fp8_format = Format.HYBRID
+        fp8_recipe = DelayedScaling(
+            fp8_format=fp8_format, amax_history_len=16, amax_compute_algo="max"
+        )
+        return te.fp8_autocast(enabled=True, fp8_recipe=fp8_recipe)
     else:
         return nullcontext()
 
@@ -87,7 +88,17 @@ def cast_precisions(
 
 
 @profile
-def preprocess_surface_data(batch):
+def preprocess_surface_data(
+    batch: dict,
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
+
+    """
+    Preprocess the surface data.  The functional input
+    is the air density and stream velocity.  The embeddings
+    are the surface mesh centers and normals.  The targets are
+    normalized to mean of 0, std 1.  We cache the mean and std
+    to de-normalize when computing the metrics.
+    """
 
     mesh_centers = batch["surface_mesh_centers"]
     normals = batch["surface_normals"]
@@ -100,6 +111,10 @@ def preprocess_surface_data(batch):
     norm_mean = targets.mean(dim=1)
     norm_std = targets.std(dim=1)
     targets = (targets - norm_mean) / norm_std
+
+    # If you want to use this, be sure to updat the
+    # functional_dim value in your configuration
+
     # fourier_sin_features = [
     #     torch.sin(mesh_centers * (2 ** i) * torch.pi)
     #     for i in range(4)
@@ -143,7 +158,16 @@ def preprocess_surface_data(batch):
     return node_features, embeddings, targets, others
 
 
-def preprocess_volume_data(batch):
+def preprocess_volume_data(
+    batch: dict,
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
+    """
+    Preprocess the volumetric data.  Right now, it's just
+    normalizing the targets and using the mesh centers as embeddings.
+
+    The targets are normalized to mean of 0, std 1.  We cache the
+    mean and std to de-normalize when computing the metrics.
+    """
 
     mesh_centers = batch["volume_mesh_centers"]
     targets = batch["volume_fields"]
@@ -154,11 +178,11 @@ def preprocess_volume_data(batch):
     embeddings = mesh_centers
 
     # Normalize the surface fields:
-    norm_mean = targets.mean(dim=0)
-    norm_std = targets.std(dim=0)
+    norm_mean = targets.mean(dim=1)
+    norm_std = targets.std(dim=1)
     targets = (targets - norm_mean) / norm_std
 
-    node_features = node_features.unsqueeze(0).broadcast_to(embeddings.shape[0], -1)
+    node_features = node_features.unsqueeze(0).broadcast_to(1, embeddings.shape[1], -1)
 
     others = {
         "norm_mean": norm_mean,
@@ -168,7 +192,17 @@ def preprocess_volume_data(batch):
 
 
 @profile
-def downsample_surface(features, embeddings, targets, num_keep=1024):
+def downsample_surface(
+    features: torch.Tensor,
+    embeddings: torch.Tensor,
+    targets: torch.Tensor,
+    num_keep=1024,
+):
+    """
+    Downsample the surface data. We generate one set of indices, and
+    use it to sample the same points from the features, embeddings,
+    and targets.  Using torch.multinomial to sample without replacement.
+    """
     # Determine the number of samples to keep (e.g., 50% of original size)
     num_samples = features.shape[1]
     # Generate random indices to keep (faster for large num_samples)
@@ -181,15 +215,23 @@ def downsample_surface(features, embeddings, targets, num_keep=1024):
     downsampled_embeddings = embeddings[:, indices]
     downsampled_targets = targets[:, indices]
 
-    # downsampled_features = downsampled_features.unsqueeze(0)
-    # downsampled_embeddings = downsampled_embeddings.unsqueeze(0)
-    # downsampled_targets = downsampled_targets.unsqueeze(0)
-
     return downsampled_features, downsampled_embeddings, downsampled_targets
 
 
 @profile
-def downsample_volume(features, embeddings, targets, num_keep=1024):
+def downsample_volume(
+    features: torch.Tensor,
+    embeddings: torch.Tensor,
+    targets: torch.Tensor,
+    num_keep=1024,
+):
+    """
+    Downsample the volume data.  torch.multinomial has a limit of 2^24
+    for num_samples, and the volumetric data typically exceeds that.
+
+    So, this isjust sampling randomly with num_keep.  The hope
+    is that the duplication is small ... but this needs to be refined.
+    """
     # Determine the number of samples to keep (e.g., 50% of original size)
     num_samples = features.shape[1]
     # The volume data is so large, that we'll sample randints
@@ -204,28 +246,81 @@ def downsample_volume(features, embeddings, targets, num_keep=1024):
     return downsampled_features, downsampled_embeddings, downsampled_targets
 
 
+def forward_pass(
+    batch: dict,
+    model: torch.nn.Module,
+    precision: str,
+    output_pad_size: int | None,
+    dist_manager: DistributedManager,
+    cfg: DictConfig,
+):
+    """
+    Run the forward pass of the model for one batch, including metrics and loss calculation.
+    """
+
+    if cfg.data.mode == "surface":
+        features, embeddings, targets, others = preprocess_surface_data(batch)
+        features, embeddings, targets = downsample_surface(
+            features, embeddings, targets, cfg.data.resolution
+        )
+
+    elif cfg.data.mode == "volume":
+        features, embeddings, targets, others = preprocess_volume_data(batch)
+        features, embeddings, targets = downsample_volume(
+            features, embeddings, targets, cfg.data.resolution
+        )
+    else:
+        raise ValueError(f"Unknown data mode: {cfg.data.mode}")
+
+    # Cast precisions:
+    features, embeddings = cast_precisions(features, embeddings, precision)
+
+    with get_autocast_context(precision):
+        outputs = model(features, embeddings)
+        if output_pad_size is not None:
+            # Remove the padded outputs:
+            outputs = outputs[:, :, :-output_pad_size]
+        loss = loss_fn(outputs, targets, others, cfg.data.mode)
+
+    metrics = metrics_fn(outputs, targets, others, dist_manager, cfg.data.mode)
+
+    return loss, metrics
+
+
 @profile
 def train_epoch(
     dataloader,
-    sampler,
-    model,
-    optimizer,
-    scheduler,
-    logger,
-    writer,
-    epoch,
-    cfg,
-    dist_manager,
-    scaler=None,  # <-- Added scaler argument
-):
-    """Train for one epoch
+    sampler: torch.utils.data.Sampler | None,
+    model: torch.nn.Module,
+    output_pad_size: int | None,
+    optimizer: torch.optim.Optimizer,
+    scheduler: torch.optim.lr_scheduler._LRScheduler,
+    logger: PythonLogger,
+    writer: SummaryWriter,
+    epoch: int,
+    cfg: DictConfig,
+    dist_manager: DistributedManager,
+    scaler: GradScaler | None = None,
+) -> float:
+    """
+    Train the model for one epoch.
 
     Args:
-        dataloader: Training data loader
-        model: The model to train
-        logger: Python logger instance
-        writer: Tensorboard writer
-        cfg: Configuration object
+        dataloader (list[dict]): Training data loader
+        sampler (torch.utils.data.Sampler | None): Sampler for distributed or sequential sampling.
+        model (torch.nn.Module): The neural network model to train.
+        output_pad_size (int | None): Optional output padding size for lowest precisions (FP8).
+        optimizer (torch.optim.Optimizer): Optimizer for model parameters.
+        scheduler (torch.optim.lr_scheduler._LRScheduler): Learning rate scheduler.
+        logger (PythonLogger): Logger for training progress.
+        writer (SummaryWriter): TensorBoard writer for logging metrics.
+        epoch (int): Current epoch number.
+        cfg (DictConfig): Hydra configuration object.
+        dist_manager (DistributedManager): Distributed manager from physicsnemo.
+        scaler (GradScaler | None, optional): Gradient scaler for mixed precision training.
+
+    Returns:
+        float: The average training loss for the epoch.
     """
     model.train()
     total_loss = 0
@@ -234,38 +329,17 @@ def train_epoch(
     epoch_indices = list(sampler) if sampler is not None else range(len(dataloader))
     epoch_len = len(epoch_indices)
     precision = getattr(cfg.training, "precision", "float32")
-    context = get_autocast_context(precision)
     start_time = time.time()
     with Profiler():
         for i, batch_idx in enumerate(epoch_indices):
-
             batch = dataloader[batch_idx]
-            # Get data from batch
-            if cfg.data.mode == "surface":
-                features, embeddings, targets, others = preprocess_surface_data(batch)
-                features, embeddings, targets = downsample_surface(
-                    features, embeddings, targets, cfg.data.resolution
-                )
-            elif cfg.data.mode == "volume":
-                features, embeddings, targets, others = preprocess_volume_data(batch)
-                features, embeddings, targets = downsample_volume(
-                    features, embeddings, targets, cfg.data.resolution
-                )
-            else:
-                raise ValueError(f"Unknown data mode: {cfg.data.mode}")
-
             # preload the next batch, if we're not on the last batch
             if i < epoch_len - 1 and sampler is not None:
                 dataloader.preload(epoch_indices[i + 1])
 
-            # Cast precisions:
-            features, embeddings = cast_precisions(features, embeddings, precision)
-
-            with context:
-                outputs = model(features, embeddings)
-                loss = loss_fn(outputs, targets, others, cfg.data.mode)
-
-            metrics = metrics_fn(outputs, targets, others, dist_manager, cfg.data.mode)
+            loss, metrics = forward_pass(
+                batch, model, precision, output_pad_size, dist_manager, cfg
+            )
 
             optimizer.zero_grad()
             if precision == "float16" and scaler is not None:
@@ -282,7 +356,8 @@ def train_epoch(
             end_time = time.time()
 
             # Logging
-            total_loss += loss.item()
+            this_loss = loss.detach().item()
+            total_loss += this_loss
 
             if i == 0:
                 total_metrics = metrics
@@ -296,7 +371,7 @@ def train_epoch(
             images_per_second = 1 / duration
 
             logger.info(
-                f"Epoch {epoch} [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
+                f"Epoch {epoch} [{i}/{epoch_len}] Loss: {this_loss:.6f} Duration: {duration:.2f}s"
             )
             if dist_manager.rank == 0:
                 writer.add_scalar(
@@ -304,7 +379,7 @@ def train_epoch(
                     optimizer.param_groups[0]["lr"],
                     i + epoch_len * epoch,
                 )
-                writer.add_scalar("batch/loss", loss.item(), i + epoch_len * epoch)
+                writer.add_scalar("batch/loss", this_loss, i + epoch_len * epoch)
                 writer.add_scalar(
                     "batch/throughpu_per_gpu", images_per_second, i + epoch_len * epoch
                 )
@@ -332,25 +407,31 @@ def train_epoch(
 @profile
 def val_epoch(
     dataloader,
-    sampler,
-    model,
-    logger,
-    val_writer,
-    epoch,
-    cfg,
-    dist_manager,
-):
-    """Validation for one epoch
+    sampler: torch.utils.data.Sampler | None,
+    model: torch.nn.Module,
+    output_pad_size: int | None,
+    logger: PythonLogger,
+    val_writer: SummaryWriter,
+    epoch: int,
+    cfg: DictConfig,
+    dist_manager: DistributedManager,
+) -> float:
+    """
+    Run validation for one epoch.
 
     Args:
-        dataloader: Validation data loader
-        sampler: Validation data sampler
-        model: The model to evaluate
-        logger: Python logger instance
-        writer: Tensorboard writer
-        epoch: Current epoch number
-        cfg: Configuration object
-        dist_manager: Distributed manager instance
+        dataloader (list[dict]): Validation data loader.
+        sampler (torch.utils.data.Sampler | None): Sampler for distributed or sequential sampling.
+        model (torch.nn.Module): The model to evaluate.
+        output_pad_size (int | None): Optional output padding size for lowest precisions (FP8).
+        logger (PythonLogger): Logger for validation progress.
+        val_writer (SummaryWriter): TensorBoard writer for logging validation metrics.
+        epoch (int): Current epoch number.
+        cfg (DictConfig): Hydra configuration object.
+        dist_manager (DistributedManager): Distributed manager instance.
+
+    Returns:
+        float: The average validation loss for the epoch.
     """
 
     model.eval()  # Set model to evaluation mode
@@ -360,31 +441,20 @@ def val_epoch(
     epoch_indices = list(sampler) if sampler is not None else range(len(dataloader))
     epoch_len = len(epoch_indices)
     precision = getattr(cfg.training, "precision", "float32")
-    context = get_autocast_context(precision)
 
     start_time = time.time()
     with torch.no_grad():  # Disable gradient computation
         for i, batch_idx in enumerate(epoch_indices):
-            batch = dataloader[batch_idx]
             # Get data from batch
-            if cfg.data.mode == "surface":
-                features, embeddings, targets, others = preprocess_surface_data(batch)
-                features, embeddings, targets = downsample_surface(
-                    features, embeddings, targets, cfg.data.resolution
-                )
-            elif cfg.data.mode == "volume":
-                features, embeddings, targets, others = preprocess_volume_data(batch)
-                features, embeddings, targets = downsample_volume(
-                    features, embeddings, targets, cfg.data.resolution
-                )
-            else:
-                raise ValueError(f"Unknown data mode: {cfg.data.mode}")
+            batch = dataloader[batch_idx]
 
-            with context:
-                outputs = model(features, embeddings)
-                loss = loss_fn(outputs, targets, others, cfg.data.mode)
+            # preload the next batch, if we're not on the last batch
+            if i < epoch_len - 1 and sampler is not None:
+                dataloader.preload(epoch_indices[i + 1])
 
-            metrics = metrics_fn(outputs, targets, others, dist_manager, cfg.data.mode)
+            loss, metrics = forward_pass(
+                batch, model, precision, output_pad_size, dist_manager, cfg
+            )
 
             if i == 0:
                 total_metrics = metrics
@@ -394,16 +464,17 @@ def val_epoch(
                 }
 
             # Logging
-            total_loss += loss.item()
+            this_loss = loss.detach().item()
+            total_loss += this_loss
 
             end_time = time.time()
             duration = end_time - start_time
             start_time = end_time
 
             logger.info(
-                f"Val [{i}/{epoch_len}] Loss: {loss.item():.6f} Duration: {duration:.2f}s"
+                f"Val [{i}/{epoch_len}] Loss: {this_loss:.6f} Duration: {duration:.2f}s"
             )
-            # We don't add individual loss measurements in the validation loop.
+            # We don't add individual loss measurements to tensorboard in the validation loop.
 
     avg_loss = total_loss / epoch_len
     avg_metrics = {k: v / epoch_len for k, v in total_metrics.items()}
@@ -454,6 +525,25 @@ def main(cfg: DictConfig):
 
     logger.info(f"Config:\n{omegaconf.OmegaConf.to_yaml(cfg, resolve=True)}")
 
+    if cfg.training.precision == "float8":
+        # we have to manipulate the output shape
+        # to enable fp8 computations with transformer_engine.
+        # need the output to be divisible by 16.
+        # if (cfg.model.embedding_dim + cfg.model.functional_dim) % 16 != 0:
+
+        if cfg.model.out_dim % 16 != 0:
+            # pad the output:
+            output_pad_size = 16 - (cfg.model.out_dim % 16)
+            cfg.model.out_dim += output_pad_size
+            logger.info(
+                f"Padding output dimension to {cfg.model.out_dim} for fp8 autocast"
+            )
+        else:
+            output_pad_size = None
+    else:
+        input_pad_size = None
+        output_pad_size = None
+
     # Set up model
     model = hydra.utils.instantiate(cfg.model)
 
@@ -564,6 +654,7 @@ def main(cfg: DictConfig):
             train_dataset,
             train_sampler,
             model,
+            output_pad_size,
             optimizer,
             scheduler,
             logger,
@@ -582,6 +673,7 @@ def main(cfg: DictConfig):
             val_dataset,
             val_sampler,
             model,
+            output_pad_size,
             logger,
             val_writer,
             epoch,

From 2332a71d931a977c38adf2c8d1c4b2188f838df0 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Mon, 28 Jul 2025 20:19:33 +0000
Subject: [PATCH 14/23] Update model to not block multipl layers per mlp

---
 physicsnemo/models/transolver/transolver.py | 36 +++++++--------------
 1 file changed, 12 insertions(+), 24 deletions(-)

diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index 51b953a2cb..129314411e 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -93,20 +93,20 @@ def __init__(
 
         self.linear_pre = linear_layer(n_input, n_hidden)
         self.linear_post = linear_layer(n_hidden, n_output)
-        # self.linears = nn.ModuleList(
-        #     [
-        #         nn.Sequential(linear_layer(n_hidden, n_hidden), act())
-        #         for _ in range(n_layers)
-        #     ]
-        # )
+        self.linears = nn.ModuleList(
+            [
+                nn.Sequential(linear_layer(n_hidden, n_hidden), act())
+                for _ in range(n_layers)
+            ]
+        )
 
     def forward(self, x):
         x = self.act(self.linear_pre(x))
-        # for i in range(self.n_layers):
-        #     if self.res:
-        #         x = self.linears[i](x) + x
-        #     else:
-        #         x = self.linears[i](x)
+        for i in range(self.n_layers):
+            if self.res:
+                x = self.linears[i](x) + x
+            else:
+                x = self.linears[i](x)
         x = self.linear_post(x)
         return x
 
@@ -180,18 +180,6 @@ def __init__(
                 hidden_size=hidden_dim,
                 ffn_hidden_size=hidden_dim * mlp_ratio,
             )
-            # self.ln_mlp1 = nn.Sequential(
-            #     te.LayerNorm(hidden_dim),
-            #     MLP(
-            #         hidden_dim,
-            #         hidden_dim * mlp_ratio,
-            #         hidden_dim,
-            #         n_layers=0,
-            #         res=False,
-            #         act=act,
-            #         use_te=True
-            #     ),
-            # )
         else:
             self.ln_mlp1 = nn.Sequential(
                 nn.LayerNorm(hidden_dim),
@@ -490,7 +478,7 @@ def get_grid(self, ref: int, batchsize: int = 1) -> torch.Tensor:
 
     def forward(
         self,
-        fx: torch.Tensor,
+        fx: torch.Tensor | None,
         embedding: torch.Tensor | None = None,
         T: torch.Tensor = None,
     ) -> torch.Tensor:

From 93d33c26fb95f9f41f4dfcaf5e4ab00c6420be8f Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Mon, 28 Jul 2025 20:29:28 +0000
Subject: [PATCH 15/23] Enabling domain parallelism.

---
 .../transolver/datapipe.py                    | 184 ++++++++-----
 .../transolver/metrics.py                     |  14 +-
 .../external_aerodynamics/transolver/train.py | 252 ++++++------------
 3 files changed, 202 insertions(+), 248 deletions(-)

diff --git a/examples/cfd/external_aerodynamics/transolver/datapipe.py b/examples/cfd/external_aerodynamics/transolver/datapipe.py
index 6d42b349f6..b1b65a1c3a 100644
--- a/examples/cfd/external_aerodynamics/transolver/datapipe.py
+++ b/examples/cfd/external_aerodynamics/transolver/datapipe.py
@@ -37,6 +37,10 @@
 
 import numpy as np
 import zarr
+import zarrs
+
+zarr.config.set({"codec_pipeline.path": "zarrs.ZarrsCodecPipeline"})
+
 
 import torch
 import torch.distributed as dist
@@ -231,6 +235,56 @@ def __del__(self):
         if hasattr(self, "executor"):
             self.executor.shutdown(wait=True)
 
+    def _get_slice_boundaries(
+        self, zarr_array: zarr.Array
+    ) -> Tuple[int, int, tuple | None]:
+        # Determine what slice this rank should read
+        if self.device_mesh is not None:
+            # How many splits to make?
+            n_splits = dist.get_world_size(group=self.device_mesh.get_group())
+            # What rank is this one?
+            this_rank = self.device_mesh.get_local_rank()
+
+            sections = compute_split_shapes(zarr_array.shape[0], n_splits)
+
+            global_chunk_start = sum(sections[:this_rank])
+            global_chunk_stop = global_chunk_start + sections[this_rank]
+
+            chunk_sizes = tuple(
+                (section,) + zarr_array.shape[1:] for section in sections
+            )
+
+        else:
+            global_chunk_start, global_chunk_stop = 0, zarr_array.shape[0]
+            chunk_sizes = None
+
+        return global_chunk_start, global_chunk_stop, chunk_sizes
+
+    def create_tensor_spec(
+        self,
+        zarr_array: zarr.Array,
+        key: str,
+        placements: tuple[Placement],
+        sharding_shapes: dict[int, tuple[int]] | None = None,
+    ) -> ShardTensorSpec:
+
+        # Unpack the batch index:
+        shape = (1,) + zarr_array.shape
+        # Don't forget to unpack it in the sharding shapes too:
+        if sharding_shapes is not None:
+            for k in sharding_shapes.keys():
+                sharding_shapes[k] = tuple((1,) + s for s in sharding_shapes[k])
+
+        stride = _stride_from_contiguous_shape_C_style(shape)
+
+        meta = TensorMeta(shape, stride, to_torch_dtype(zarr_array.dtype))
+        return ShardTensorSpec(
+            mesh=self.device_mesh,
+            placements=placements,
+            tensor_meta=meta,
+            _sharding_shapes=sharding_shapes,
+        )
+
     @profile
     def _read_key_chunk_aligned(
         self, zarr_group: zarr.Group, key: str, futures: list[Future]
@@ -250,34 +304,14 @@ def _read_key_chunk_aligned(
 
         zarr_array = zarr_group[key]
 
-        # Determine what slice this rank should read
-        if self.device_mesh is not None:
-            # How many splits to make?
-            n_splits = dist.get_world_size(group=self.device_mesh.get_group())
-            # What rank is this one?
-            this_rank = self.device_mesh.get_local_rank()
-
-            sections = compute_split_shapes(zarr_array.shape[0], n_splits)
-
-            global_chunk_start = sum(sections[:this_rank])
-            global_chunk_stop = global_chunk_start + sections[this_rank]
-
-            chunk_sizes = tuple(
-                (section,) + zarr_array.shape[1:] for section in sections
-            )
-            stride = _stride_from_contiguous_shape_C_style(zarr_array.shape)
-
-            meta = TensorMeta(zarr_array.shape, stride, zarr_array.dtype)
-            self.tensor_specs[key] = ShardTensorSpec(
-                mesh=self.device_mesh,
-                placements=(Shard(0),),
-                tensor_meta=meta,
-                _sharding_shapes={0: chunk_sizes},
+        global_chunk_start, global_chunk_stop, chunk_sizes = self._get_slice_boundaries(
+            zarr_array
+        )
+        if chunk_sizes is not None:
+            self.tensor_specs[key] = self.create_tensor_spec(
+                zarr_array, key, (Shard(1),), {0: chunk_sizes}
             )
 
-        else:
-            global_chunk_start, global_chunk_stop = 0, zarr_array.shape[0]
-
         # Calculate the shape of data this rank will read
         local_shape = [global_chunk_stop - global_chunk_start] + list(
             zarr_array.shape[1:]
@@ -374,11 +408,10 @@ def _read_key_standard(
         if zarr_array.shape == ():
 
             if self.device_mesh is not None:
-                self.tensor_specs[key] = ShardTensorSpec(
-                    mesh=self.device_mesh,
-                    placements=(Replicate(),),
-                    tensor_meta=TensorMeta(zarr_array.shape, (), zarr_array.dtype),
-                    _sharding_shapes={},
+                self.tensor_specs[key] = self.create_tensor_spec(
+                    zarr_array,
+                    key,
+                    (Replicate(),),
                 )
 
             output = torch.from_numpy(np.array(zarr_array))
@@ -386,34 +419,44 @@ def _read_key_standard(
                 output = output.pin_memory()
             return output
 
+        global_chunk_start, global_chunk_stop, chunk_sizes = self._get_slice_boundaries(
+            zarr_array
+        )
+        if chunk_sizes is not None:
+            self.tensor_specs[key] = self.create_tensor_spec(
+                zarr_array, key, (Shard(1),), {0: chunk_sizes}
+            )
+
+        # Calculate the shape of data this rank will read
+        local_shape = [global_chunk_stop - global_chunk_start] + list(
+            zarr_array.shape[1:]
+        )
+
         # data = np.empty(zarr_array.shape, dtype=zarr_array.dtype)
         output = torch.empty(
-            zarr_array.shape,
+            local_shape,
             dtype=to_torch_dtype(zarr_array.dtype),
             pin_memory=self.pin_memory,
         )
         data = output.numpy()
         slice = np.s_[:]
+
+        # The zarr slice is not the numpy slice if we're sharding
+        if chunk_sizes is not None:
+            zarr_slice = np.s_[global_chunk_start:global_chunk_stop]
+        else:
+            zarr_slice = np.s_[:]
+
         futures.append(
             self.executor.submit(
                 _read_chunk_into_array,
                 data,
                 zarr_array,
                 slice,
+                zarr_slice,
             )
         )
 
-        if self.device_mesh is not None:
-            # We need to track the tensor meta for this key
-            stride = _stride_from_contiguous_shape_C_style(zarr_array.shape)
-
-            self.tensor_specs[key] = ShardTensorSpec(
-                mesh=self.device_mesh,
-                placements=(Replicate(),),
-                tensor_meta=TensorMeta(zarr_array.shape, stride, zarr_array.dtype),
-                _sharding_shapes={},
-            )
-
         return output
 
     @profile
@@ -426,28 +469,30 @@ def _read_zarr_file(self, filepath: Path) -> Dict[str, np.ndarray]:
         Returns:
             Dictionary mapping keys to numpy arrays or torch tensors.
         """
-        with zarr.open_group(filepath, mode="r") as zarr_group:
-            data = {}
-            futures = []
+        zarr_group = zarr.open_group(str(filepath), mode="r")
+        # group_keys = list(zarr_group.keys())
 
-            # Process each key
-            for key in self.keys_to_read:
+        data = {}
+        futures = []
 
-                if key not in zarr_group:
-                    continue
+        # Process each key
+        for key in self.keys_to_read:
 
-                if key in self.large_keys:
-                    # Use chunk-aligned reading for large data
-                    data[key] = self._read_key_chunk_aligned(zarr_group, key, futures)
-                else:
-                    # Use simple reading for other data
-                    data[key] = self._read_key_standard(zarr_group, key, futures)
+            # if key not in group_keys:
+            #     continue
+
+            if key in self.large_keys:
+                # Use chunk-aligned reading for large data
+                data[key] = self._read_key_chunk_aligned(zarr_group, key, futures)
+            else:
+                # Use simple reading for other data
+                data[key] = self._read_key_standard(zarr_group, key, futures)
 
-            # Wait for all futures to complete
-            for future in futures:
-                future.result()
+        # Wait for all futures to complete
+        for future in futures:
+            future.result()
 
-            return data
+        return data
 
     @profile
     def _move_to_gpu(self, data: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]:
@@ -461,16 +506,6 @@ def _move_to_gpu(self, data: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]
         """
         result = {}
 
-        # This is old code.
-        # Thinking about ways to reduce CPU overhead.
-        # If we could avoid the data reading step, it would be easier.
-        # total_size = 0
-        # for key, array in data.items():
-        #     size = array.numel() * array.element_size()
-        #     total_size += size
-        #     print(f"Size in memory of {key}: {size / 1024 / 1024 / 1024} GB")
-        # print(f"Total size in memory: {total_size / 1024 / 1024 / 1024} GB")
-
         with torch.cuda.stream(self.data_loader_stream):
 
             for key, array in data.items():
@@ -510,10 +545,13 @@ def _convert_to_shard_tensors(
             )
 
             # Find out the desired placement:
-            if isinstance(self.placements, dict):
-                target_placement = self.placements[key]
+            if tensor.numel() > 1:
+                if isinstance(self.placements, dict):
+                    target_placement = self.placements[key]
+                else:
+                    target_placement = self.placements
             else:
-                target_placement = self.placements
+                target_placement = (Replicate(),)
 
             # Redistribute if necessary:
             # (Recall that this is one dimensional mesh only)
diff --git a/examples/cfd/external_aerodynamics/transolver/metrics.py b/examples/cfd/external_aerodynamics/transolver/metrics.py
index 18e8ecfa2e..5275710f06 100644
--- a/examples/cfd/external_aerodynamics/transolver/metrics.py
+++ b/examples/cfd/external_aerodynamics/transolver/metrics.py
@@ -16,7 +16,7 @@
 
 import torch
 import torch.distributed as dist
-
+from physicsnemo.distributed import ShardTensor
 from physicsnemo.distributed import DistributedManager
 
 
@@ -39,8 +39,16 @@ def all_reduce_dict(
         return metrics
 
     for key, value in metrics.items():
-        dist.all_reduce(value)
-        value = value / dm.world_size
+        if isinstance(value, ShardTensor):
+            # Perform the reduction over the ddp axis, not the domain:
+            value = value.full_tensor()
+            mesh = dm.global_mesh["ddp"]
+            dist.all_reduce(value, group=mesh.get_group())
+            value = value / mesh.size()
+        else:
+            dist.all_reduce(value)
+            value = value / dm.world_size
+
         metrics[key] = value
 
     return metrics
diff --git a/examples/cfd/external_aerodynamics/transolver/train.py b/examples/cfd/external_aerodynamics/transolver/train.py
index 954dd42a2d..0a7702845a 100644
--- a/examples/cfd/external_aerodynamics/transolver/train.py
+++ b/examples/cfd/external_aerodynamics/transolver/train.py
@@ -23,6 +23,14 @@
 from omegaconf import DictConfig
 from torch.utils.tensorboard import SummaryWriter
 
+
+from torch.distributed.fsdp import (
+    FullyShardedDataParallel as FSDP,
+    ShardingStrategy,
+)
+from torch.distributed.tensor.placement_types import Shard
+from torch.distributed.tensor import distribute_module
+
 from physicsnemo.launch.utils import load_checkpoint, save_checkpoint
 from physicsnemo.launch.logging import PythonLogger, RankZeroLoggingWrapper
 from physicsnemo.distributed import DistributedManager
@@ -33,6 +41,12 @@
 from datapipe import DomainParallelZarrDataset
 from loss import loss_fn
 from metrics import metrics_fn
+from preprocess import (
+    preprocess_volume_data,
+    preprocess_surface_data,
+    downsample_volume,
+    downsample_surface,
+)
 
 from contextlib import nullcontext
 from torch.amp import autocast, GradScaler
@@ -87,165 +101,6 @@ def cast_precisions(
         return features, embeddings
 
 
-@profile
-def preprocess_surface_data(
-    batch: dict,
-) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
-
-    """
-    Preprocess the surface data.  The functional input
-    is the air density and stream velocity.  The embeddings
-    are the surface mesh centers and normals.  The targets are
-    normalized to mean of 0, std 1.  We cache the mean and std
-    to de-normalize when computing the metrics.
-    """
-
-    mesh_centers = batch["surface_mesh_centers"]
-    normals = batch["surface_normals"]
-    targets = batch["surface_fields"]
-    node_features = torch.stack(
-        [batch["air_density"], batch["stream_velocity"]], dim=-1
-    ).to(torch.float32)
-
-    # Normalize the surface fields:
-    norm_mean = targets.mean(dim=1)
-    norm_std = targets.std(dim=1)
-    targets = (targets - norm_mean) / norm_std
-
-    # If you want to use this, be sure to updat the
-    # functional_dim value in your configuration
-
-    # fourier_sin_features = [
-    #     torch.sin(mesh_centers * (2 ** i) * torch.pi)
-    #     for i in range(4)
-    # ]
-    # fourier_cos_features = [
-    #     torch.cos(mesh_centers * (2 ** i) * torch.pi)
-    #     for i in range(4)
-    # ]
-
-    # Calculate center of mass
-    sizes = batch["stl_areas"]
-    centers = batch["stl_centers"]
-
-    total_weighted_position = torch.einsum("ki,kij->kj", sizes, centers)
-    total_size = torch.sum(sizes)
-    center_of_mass = total_weighted_position[None, ...] / total_size
-
-    # Subtract the COM from the centers:
-    mesh_centers = mesh_centers - center_of_mass
-
-    embeddings = torch.cat(
-        [
-            mesh_centers,
-            normals,
-            # *fourier_sin_features,
-            # *fourier_cos_features
-        ],
-        dim=-1,
-    )
-    node_features = node_features.unsqueeze(1).broadcast_to(1, embeddings.shape[1], -1)
-
-    others = {
-        "surface_areas": sizes,
-        "surface_normals": normals,
-        "stream_velocity": batch["stream_velocity"],
-        "air_density": batch["air_density"],
-        "norm_mean": norm_mean,
-        "norm_std": norm_std,
-    }
-
-    return node_features, embeddings, targets, others
-
-
-def preprocess_volume_data(
-    batch: dict,
-) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
-    """
-    Preprocess the volumetric data.  Right now, it's just
-    normalizing the targets and using the mesh centers as embeddings.
-
-    The targets are normalized to mean of 0, std 1.  We cache the
-    mean and std to de-normalize when computing the metrics.
-    """
-
-    mesh_centers = batch["volume_mesh_centers"]
-    targets = batch["volume_fields"]
-    node_features = torch.stack(
-        [batch["air_density"], batch["stream_velocity"]], dim=-1
-    ).to(torch.float32)
-
-    embeddings = mesh_centers
-
-    # Normalize the surface fields:
-    norm_mean = targets.mean(dim=1)
-    norm_std = targets.std(dim=1)
-    targets = (targets - norm_mean) / norm_std
-
-    node_features = node_features.unsqueeze(0).broadcast_to(1, embeddings.shape[1], -1)
-
-    others = {
-        "norm_mean": norm_mean,
-        "norm_std": norm_std,
-    }
-    return node_features, embeddings, targets, others
-
-
-@profile
-def downsample_surface(
-    features: torch.Tensor,
-    embeddings: torch.Tensor,
-    targets: torch.Tensor,
-    num_keep=1024,
-):
-    """
-    Downsample the surface data. We generate one set of indices, and
-    use it to sample the same points from the features, embeddings,
-    and targets.  Using torch.multinomial to sample without replacement.
-    """
-    # Determine the number of samples to keep (e.g., 50% of original size)
-    num_samples = features.shape[1]
-    # Generate random indices to keep (faster for large num_samples)
-    indices = torch.multinomial(
-        torch.ones(num_samples, device=features.device), num_keep, replacement=False
-    )
-
-    # Use the same indices to downsample all tensors
-    downsampled_features = features[:, indices]
-    downsampled_embeddings = embeddings[:, indices]
-    downsampled_targets = targets[:, indices]
-
-    return downsampled_features, downsampled_embeddings, downsampled_targets
-
-
-@profile
-def downsample_volume(
-    features: torch.Tensor,
-    embeddings: torch.Tensor,
-    targets: torch.Tensor,
-    num_keep=1024,
-):
-    """
-    Downsample the volume data.  torch.multinomial has a limit of 2^24
-    for num_samples, and the volumetric data typically exceeds that.
-
-    So, this isjust sampling randomly with num_keep.  The hope
-    is that the duplication is small ... but this needs to be refined.
-    """
-    # Determine the number of samples to keep (e.g., 50% of original size)
-    num_samples = features.shape[1]
-    # The volume data is so large, that we'll sample randints
-    # which will very rarely duplicate
-    indices = torch.randint(0, num_samples, (num_keep,), device=features.device)
-
-    # Use the same indices to downsample all tensors
-    downsampled_features = features[:, indices]
-    downsampled_embeddings = embeddings[:, indices]
-    downsampled_targets = targets[:, indices]
-
-    return downsampled_features, downsampled_embeddings, downsampled_targets
-
-
 def forward_pass(
     batch: dict,
     model: torch.nn.Module,
@@ -274,8 +129,17 @@ def forward_pass(
 
     # Cast precisions:
     features, embeddings = cast_precisions(features, embeddings, precision)
-
     with get_autocast_context(precision):
+        print(
+            f"features shape: {features.shape} nd placements: {features._spec.placements}"
+        )
+        print(
+            f"embeddings shape: {embeddings.shape} nd placements: {embeddings._spec.placements}"
+        )
+        print(
+            f"targets shape: {targets.shape} nd placements: {targets._spec.placements}"
+        )
+        print(f"cat shape: {torch.cat((features, embeddings), dim=-1).shape}")
         outputs = model(features, embeddings)
         if output_pad_size is not None:
             # Remove the padded outputs:
@@ -332,6 +196,9 @@ def train_epoch(
     start_time = time.time()
     with Profiler():
         for i, batch_idx in enumerate(epoch_indices):
+
+            if i > 10:
+                break
             batch = dataloader[batch_idx]
             # preload the next batch, if we're not on the last batch
             if i < epoch_len - 1 and sampler is not None:
@@ -445,6 +312,8 @@ def val_epoch(
     start_time = time.time()
     with torch.no_grad():  # Disable gradient computation
         for i, batch_idx in enumerate(epoch_indices):
+            if i > 10:
+                break
             # Get data from batch
             batch = dataloader[batch_idx]
 
@@ -549,7 +418,39 @@ def main(cfg: DictConfig):
 
     model.to(dist_manager.device)
 
-    if dist_manager.world_size > 1:
+    # Configure domain parallelism, if enabled:
+    domain_size = int(cfg.training.domain_parallelism)
+
+    if domain_size > 1:
+        # You can use -1 to one axis to indicate that you want to use all the GPUs in that dimension.
+        mesh = dist_manager.initialize_mesh(
+            mesh_shape=(-1, domain_size), mesh_dim_names=("ddp", "domain")
+        )
+        # This is a subset of all the GPUs, and will vary depending on the process.
+        # Think of this as slicing the global mesh along the domain axis.
+        # It will contain only the GPUs that this process is sharing data with.
+        domain_mesh = mesh["domain"]
+        ddp_mesh = mesh["ddp"]
+        placements = (Shard(1),)
+    else:
+        domain_mesh = None
+        ddp_mesh = None
+        placements = None
+
+    if domain_size > 1:
+        # Instead of DDP, for sharding we use FSDP.  It's possible to use FSDP in the DDP
+        # mode, but since it's not pure data parallel we have to me more careful.
+
+        # First, distribute the model so that each GPU has the copy with DTensor weights:
+        model = distribute_module(model, domain_mesh)
+
+        model = FSDP(
+            model,
+            device_mesh=mesh["ddp"],
+            sharding_strategy=ShardingStrategy.NO_SHARD,
+        )
+
+    elif dist_manager.world_size > 1:
         model = torch.nn.parallel.DistributedDataParallel(
             model,
             device_ids=[dist_manager.local_rank],
@@ -559,15 +460,11 @@ def main(cfg: DictConfig):
     num_params = sum(p.numel() for p in model.parameters())
     logger.info(f"Number of parameters: {num_params}")
 
-    # Set up data
-    device_mesh = None
-    placements = None
-
     # Training dataset
 
     train_dataset = DomainParallelZarrDataset(
         data_path=cfg.data.train.data_path,
-        device_mesh=device_mesh,
+        device_mesh=domain_mesh,
         placements=placements,
         max_workers=cfg.data.max_workers,
         pin_memory=cfg.data.pin_memory,
@@ -579,7 +476,7 @@ def main(cfg: DictConfig):
 
     val_dataset = DomainParallelZarrDataset(
         data_path=cfg.data.val.data_path,  # Assuming validation data path is configured
-        device_mesh=device_mesh,
+        device_mesh=domain_mesh,
         placements=placements,
         max_workers=cfg.data.max_workers,
         pin_memory=cfg.data.pin_memory,
@@ -587,23 +484,34 @@ def main(cfg: DictConfig):
         large_keys=cfg.data.large_keys,
     )
 
+    if ddp_mesh is not None:
+        num_replicas = ddp_mesh.size()
+        data_rank = ddp_mesh.get_local_rank()
+    else:
+        num_replicas = dist_manager.world_size
+        data_rank = dist_manager.rank
+
     # Set up distributed samplers
     train_sampler = torch.utils.data.distributed.DistributedSampler(
         train_dataset,
-        num_replicas=dist_manager.world_size,
-        rank=dist_manager.rank,
+        num_replicas=num_replicas,
+        rank=data_rank,
         shuffle=True,
         drop_last=True,
     )
 
     val_sampler = torch.utils.data.distributed.DistributedSampler(
         val_dataset,
-        num_replicas=dist_manager.world_size,
-        rank=dist_manager.rank,
+        num_replicas=num_replicas,
+        rank=data_rank,
         shuffle=False,  # No shuffling for validation
         drop_last=True,
     )
 
+    print(
+        f"num_replicas: {num_replicas}, data_rank: {data_rank}, val targets: {list(val_sampler)}"
+    )
+
     # Set up optimizer and scheduler
     optimizer = hydra.utils.instantiate(cfg.optimizer, params=model.parameters())
 
@@ -707,7 +615,7 @@ def launch(cfg: DictConfig):
     """
     profiler = Profiler()
     # profiler.enable("torch")
-    # profiler.enable("line_profiler")
+    profiler.enable("line_profiler")
     profiler.initialize()
     main(cfg)
     profiler.finalize()

From 0a1082c1d52dd2788149e7650b4556ae354fd176 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Mon, 28 Jul 2025 20:29:58 +0000
Subject: [PATCH 16/23] Isolate preprocess steps.

---
 .../transolver/preprocess.py                  | 282 ++++++++++++++++++
 1 file changed, 282 insertions(+)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/preprocess.py

diff --git a/examples/cfd/external_aerodynamics/transolver/preprocess.py b/examples/cfd/external_aerodynamics/transolver/preprocess.py
new file mode 100644
index 0000000000..0121807e06
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/preprocess.py
@@ -0,0 +1,282 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import torch
+
+from physicsnemo.distributed.shard_tensor import ShardTensor
+from physicsnemo.utils.profiling import profile
+
+
+@profile
+def preprocess_surface_data(
+    batch: dict,
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
+
+    """
+    Preprocess the surface data.  The functional input
+    is the air density and stream velocity.  The embeddings
+    are the surface mesh centers and normals.  The targets are
+    normalized to mean of 0, std 1.  We cache the mean and std
+    to de-normalize when computing the metrics.
+    """
+
+    mesh_centers = batch["surface_mesh_centers"]
+    normals = batch["surface_normals"]
+    targets = batch["surface_fields"]
+    node_features = torch.stack(
+        [batch["air_density"], batch["stream_velocity"]], dim=-1
+    ).to(torch.float32)
+
+    # Normalize the surface fields:
+    norm_mean = targets.mean(dim=1)
+    norm_std = torch.sqrt(torch.mean((targets - norm_mean.unsqueeze(1)) ** 2, dim=1))
+    targets = (targets - norm_mean) / norm_std
+
+    # If you want to use this, be sure to updat the
+    # functional_dim value in your configuration
+
+    # fourier_sin_features = [
+    #     torch.sin(mesh_centers * (2 ** i) * torch.pi)
+    #     for i in range(4)
+    # ]
+    # fourier_cos_features = [
+    #     torch.cos(mesh_centers * (2 ** i) * torch.pi)
+    #     for i in range(4)
+    # ]
+
+    # Calculate center of mass
+    sizes = batch["stl_areas"]
+    centers = batch["stl_centers"]
+
+    total_weighted_position = torch.einsum("ki,kij->kj", sizes, centers)
+    total_size = torch.sum(sizes)
+    center_of_mass = total_weighted_position[None, ...] / total_size
+
+    # Subtract the COM from the centers:
+    mesh_centers = mesh_centers - center_of_mass
+
+    embeddings = torch.cat(
+        [
+            mesh_centers,
+            normals,
+            # *fourier_sin_features,
+            # *fourier_cos_features
+        ],
+        dim=-1,
+    )
+    # node_features = node_features.unsqueeze(1).broadcast_to(1, embeddings.shape[1], -1)
+
+    others = {
+        "surface_areas": sizes,
+        "surface_normals": normals,
+        "stream_velocity": batch["stream_velocity"],
+        "air_density": batch["air_density"],
+        "norm_mean": norm_mean,
+        "norm_std": norm_std,
+    }
+    print(
+        f"node_features shape: {node_features.shape} nd placements: {node_features._spec.placements}"
+    )
+    print(
+        f"embeddings shape: {embeddings.shape} nd placements: {embeddings._spec.placements}"
+    )
+    print(f"targets shape: {targets.shape} nd placements: {targets._spec.placements}")
+    return node_features, embeddings, targets, others
+
+
+def preprocess_volume_data(
+    batch: dict,
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
+    """
+    Preprocess the volumetric data.  Right now, it's just
+    normalizing the targets and using the mesh centers as embeddings.
+
+    The targets are normalized to mean of 0, std 1.  We cache the
+    mean and std to de-normalize when computing the metrics.
+    """
+
+    mesh_centers = batch["volume_mesh_centers"]
+    sdf = batch["sdf_volume"]
+    targets = batch["volume_fields"]
+    node_features = torch.stack(
+        [batch["air_density"], batch["stream_velocity"]], dim=-1
+    ).to(torch.float32)
+
+    # SDF is missing a trailing dimension:
+    sdf = sdf.unsqueeze(-1)
+
+    embeddings = torch.cat([mesh_centers, sdf], dim=-1)
+
+    # Normalize the surface fields:
+    norm_mean = targets.mean(dim=1)
+    # norm_std = targets.std(dim=1)
+    norm_std = torch.sqrt(torch.mean((targets - norm_mean.unsqueeze(1)) ** 2, dim=1))
+
+    targets = (targets - norm_mean) / norm_std
+
+    others = {
+        "norm_mean": norm_mean,
+        "norm_std": norm_std,
+        "stream_velocity": batch["stream_velocity"],
+        "air_density": batch["air_density"],
+    }
+    return node_features, embeddings, targets, others
+
+
+@profile
+def downsample_surface(
+    features: torch.Tensor,
+    embeddings: torch.Tensor,
+    targets: torch.Tensor,
+    num_keep=1024,
+):
+    """
+    Downsample the surface data. We generate one set of indices, and
+    use it to sample the same points from the features, embeddings,
+    and targets.  Using torch.multinomial to sample without replacement.
+    """
+    # Determine the number of samples to keep (e.g., 50% of original size)
+
+    if isinstance(features, ShardTensor):
+        local_features = features.to_local()
+        local_embeddings = embeddings.to_local()
+        local_targets = targets.to_local()
+
+        num_samples = local_embeddings.shape[1]
+
+        indices = torch.multinomial(
+            torch.ones(num_samples, device=features.device), num_keep, replacement=False
+        )
+        downsampled_embeddings = local_embeddings[:, indices]
+        downsampled_targets = local_targets[:, indices]
+        downsampled_features = features.unsqueeze(1).expand(
+            1, downsampled_embeddings.shape[1], -1
+        )
+
+        # Compute sharding shapes, replacing the local tensor num points with num_keep
+        output_embedding_shapes = {}
+        for k, shapes_tuple in embeddings._spec.sharding_shapes().items():
+            updated_shape = tuple((s[0], num_keep, s[2:]) for s in shapes_tuple)
+            output_embedding_shapes[k] = updated_shape
+
+        # Push the sampled versions back to shard tensors:
+        downsampled_embeddings = ShardTensor.from_local(
+            downsampled_embeddings,
+            device_mesh=embeddings._spec.mesh,
+            placements=embeddings._spec.placements,
+            sharding_shapes=output_embedding_shapes,
+        )
+        downsampled_targets = ShardTensor.from_local(
+            downsampled_targets,
+            device_mesh=targets._spec.mesh,
+            placements=targets._spec.placements,
+        )
+        downsampled_features = ShardTensor.from_local(
+            downsampled_features,
+            device_mesh=features._spec.mesh,
+            placements=embeddings._spec.placements,  # NOTE!  Changing the features placements here.
+        )
+
+    else:
+
+        num_samples = embeddings.shape[1]
+        # Generate random indices to keep (faster for large num_samples)
+        indices = torch.multinomial(
+            torch.ones(num_samples, device=features.device), num_keep, replacement=False
+        )
+
+        # Use the same indices to downsample all tensors
+        downsampled_embeddings = embeddings[:, indices]
+        downsampled_targets = targets[:, indices]
+        downsampled_features = features.unsqueeze(1).expand(
+            1, downsampled_embeddings.shape[1], -1
+        )
+
+    return downsampled_features, downsampled_embeddings, downsampled_targets
+
+
+@profile
+def downsample_volume(
+    features: torch.Tensor,
+    embeddings: torch.Tensor,
+    targets: torch.Tensor,
+    num_keep=1024,
+):
+    """
+    Downsample the volume data.  torch.multinomial has a limit of 2^24
+    for num_samples, and the volumetric data typically exceeds that.
+
+    So, this isjust sampling randomly with num_keep.  The hope
+    is that the duplication is small ... but this needs to be refined.
+    """
+
+    if isinstance(features, ShardTensor):
+        # For shard tensors, we slice an approximately equal amount per device
+
+        local_embeddings = embeddings.to_local()
+        local_targets = targets.to_local()
+        local_features = features.to_local()
+
+        # Make a selection of num_keep per shard:
+        local_indices = torch.randint(
+            0, local_embeddings.shape[1], (num_keep,), device=local_embeddings.device
+        )
+        downsampled_embeddings = local_embeddings[:, local_indices]
+        downsampled_targets = local_targets[:, local_indices]
+        downsampled_features = local_features.unsqueeze(1).expand(
+            1, downsampled_embeddings.shape[1], -1
+        )
+
+        # Compute sharding shapes, replacing the local tensor num points with num_keep
+        output_embedding_shapes = {}
+        for k, shapes_tuple in embeddings._spec.sharding_shapes().items():
+            updated_shape = tuple((s[0], num_keep, s[2:]) for s in shapes_tuple)
+            output_embedding_shapes[k] = updated_shape
+
+        # Push the sampled versions back to shard tensors:
+        downsampled_embeddings = ShardTensor.from_local(
+            downsampled_embeddings,
+            device_mesh=embeddings._spec.mesh,
+            placements=embeddings._spec.placements,
+            sharding_shapes=output_embedding_shapes,
+        )
+        downsampled_targets = ShardTensor.from_local(
+            downsampled_targets,
+            device_mesh=targets._spec.mesh,
+            placements=targets._spec.placements,
+        )
+        downsampled_features = ShardTensor.from_local(
+            downsampled_features,
+            device_mesh=features._spec.mesh,
+            placements=embeddings._spec.placements,  # NOTE!  Changing the features placements here.
+        )
+
+    else:
+        # Determine the number of samples to keep (e.g., 50% of original size)
+        num_samples = features.shape[1]
+        # The volume data is so large, that we'll sample randints
+        # which will very rarely duplicate
+        indices = torch.randint(0, num_samples, (num_keep,), device=features.device)
+
+        # Use the same indices to downsample all tensors
+        downsampled_embeddings = embeddings[:, indices]
+        downsampled_targets = targets[:, indices]
+
+        downsampled_features = features.unsqueeze(1).expand(
+            1, downsampled_embeddings.shape[1], -1
+        )
+
+    return downsampled_features, downsampled_embeddings, downsampled_targets

From 37d66e8e8ba07a858a70553d993c6b746380e29e Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Wed, 30 Jul 2025 09:56:05 -0700
Subject: [PATCH 17/23] Fix bug in transolver physics attention base: the wrong
 normalization was used when projecting back onto output states.

---
 physicsnemo/models/transolver/Physics_Attention.py | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index 556ea28ec3..861b1da308 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -138,15 +138,16 @@ def compute_slices_from_projections(
 
         # Cast to the computation type (since the parameter is probably fp32)
         slice_weights = slice_weights.to(slice_projections.dtype)
-        # Average the slices over the token dimension
-        slice_norm = slice_weights.sum(2)  # [Batch, N_heads, Slice_num]
+
         # This does the projection of the latent space fx by the weights:
 
         # Computing the slice tokens is a matmul followed by a normalization.
         # It can, unfortunately, overflow in reduced precision, so normalize first:
-        slice_weights = slice_weights / (slice_norm[:, :, None, :] + 1e-2)
-        slice_token = torch.matmul(slice_weights.transpose(2, 3), fx)
+        slice_norm = slice_weights.sum(2)  # [Batch, N_heads, Slice_num]
+        normed_weights = slice_weights / (slice_norm[:, :, None, :] + 1e-2)
+        slice_token = torch.matmul(normed_weights.transpose(2, 3), fx)
 
+        # Return the original weights, not the normed weights:
         return slice_weights, slice_token
 
     def compute_slice_attention(self, slice_tokens: torch.Tensor) -> torch.Tensor:
@@ -261,6 +262,8 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
         # Deslice:
         outputs = self.project_attention_outputs(out_slice_token, slice_weights)
 
+        # print(f"outputs mean and shape: {outputs.mean()} and {outputs.shape}")
+
         # Outputs now has the same shape as the original input x
 
         return outputs

From 8b61a37302d500fa72b3f0df07d1e34c59d97e90 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Thu, 31 Jul 2025 11:24:09 -0700
Subject: [PATCH 18/23] Add script for computing normalization factors for
 transolver.

---
 .../transolver/compute_normalizations.py      | 132 ++++++++++++++++++
 1 file changed, 132 insertions(+)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/compute_normalizations.py

diff --git a/examples/cfd/external_aerodynamics/transolver/compute_normalizations.py b/examples/cfd/external_aerodynamics/transolver/compute_normalizations.py
new file mode 100644
index 0000000000..b01cd6a618
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/compute_normalizations.py
@@ -0,0 +1,132 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: MIT License
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+import os
+import numpy as np
+import torch
+from omegaconf import OmegaConf
+
+from datapipe import DomainParallelZarrDataset
+
+"""
+This file provides utilities to compute normalization statistics (mean, std, min, max)
+for a given field in a dataset, typically used for preprocessing in CFD workflows.
+"""
+
+
+def compute_mean_std_min_max(
+    dataset: DomainParallelZarrDataset, field_key: str
+) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
+    """
+    Compute the mean, standard deviation, minimum, and maximum for a specified field
+    across all samples in a dataset.
+
+    Uses a numerically stable online algorithm for mean and variance.
+
+    Args:
+        dataset (DomainParallelZarrDataset): The dataset to process.
+        field_key (str): The key for the field to normalize.
+
+    Returns:
+        tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
+            mean, std, min, max tensors for the field.
+    """
+    N = 0  # Total number of elements processed
+    mean = None
+    M2 = None  # Sum of squares of differences from the current mean
+    min_val = None
+    max_val = None
+
+    for i in range(len(dataset)):
+        print(f"reading file: {i}")
+        data = dataset[i][field_key]
+        if mean is None:
+            # Initialize accumulators based on the shape of the data
+            mean = torch.zeros(data.shape[-1], device=data.device)
+            M2 = torch.zeros(data.shape[-1], device=data.device)
+            min_val = torch.full((data.shape[-1],), float("inf"), device=data.device)
+            max_val = torch.full((data.shape[-1],), float("-inf"), device=data.device)
+        n = data.shape[1]
+        N += n
+
+        # Compute batch statistics
+        batch_mean = data.mean(axis=(0, 1))
+        batch_M2 = ((data - batch_mean) ** 2).sum(axis=(0, 1))
+        batch_n = data.shape[1]
+
+        # Update min/max
+        batch_min = data.amin(dim=(0, 1))
+        batch_max = data.amax(dim=(0, 1))
+        min_val = torch.minimum(min_val, batch_min)
+        max_val = torch.maximum(max_val, batch_max)
+
+        # Update running mean and M2 (Welford's algorithm)
+        delta = batch_mean - mean
+        N += batch_n
+        mean = mean + delta * (batch_n / N)
+        M2 = M2 + batch_M2 + delta**2 * (batch_n * N) / N
+
+    var = M2 / (N - 1)
+    std = torch.sqrt(var)
+    return mean, std, min_val, max_val
+
+
+if __name__ == "__main__":
+    """
+    Script entry point for computing normalization statistics for a specified field
+    in a dataset, using configuration from a YAML file.
+
+    The computed statistics are printed and saved to a .npz file.
+    """
+    # Edit this path to your config file or hardcode the config as needed
+    config_path: str = "conf/train_surface.yaml"
+    cfg = OmegaConf.load(config_path)
+
+    # Choose which field to normalize
+    field_key: str = "surface_fields"
+
+    # Create the dataset using configuration parameters
+    dataset = DomainParallelZarrDataset(
+        data_path=cfg.data.train.data_path,
+        device_mesh=None,
+        placements=None,
+        max_workers=cfg.data.max_workers,
+        pin_memory=cfg.data.pin_memory,
+        keys_to_read=[field_key],
+        large_keys=[field_key],
+    )
+
+    # Compute normalization statistics
+    mean, std, min_val, max_val = compute_mean_std_min_max(dataset, field_key)
+    print(f"Mean for {field_key}: {mean}")
+    print(f"Std for {field_key}: {std}")
+    print(f"Min for {field_key}: {min_val}")
+    print(f"Max for {field_key}: {max_val}")
+
+    # Save statistics to a .npz file for later use
+    np.savez(
+        f"{field_key}_normalization.npz",
+        mean=mean.cpu().numpy(),
+        std=std.cpu().numpy(),
+        min=min_val.cpu().numpy(),
+        max=max_val.cpu().numpy(),
+    )

From 0253bf6a7b9314d1b19735abfbd20ee9ee4e588c Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Thu, 31 Jul 2025 11:25:10 -0700
Subject: [PATCH 19/23] remove volume config for now.

---
 .../transolver/conf/train_volume.yaml         | 92 -------------------
 1 file changed, 92 deletions(-)
 delete mode 100644 examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml

diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
deleted file mode 100644
index 97f24b9134..0000000000
--- a/examples/cfd/external_aerodynamics/transolver/conf/train_volume.yaml
+++ /dev/null
@@ -1,92 +0,0 @@
-# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
-# SPDX-FileCopyrightText: All rights reserved.
-# SPDX-License-Identifier: Apache-2.0
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-
-
-output_dir: "runs"
-run_id: "volume/OneCycleLR/noTE/bfloat16/"
-
-# Training configuration
-training:
-  precision: bfloat16 # float32, float16, bfloat16
-  num_epochs: 250
-  save_interval: 25  # Save checkpoint every N epochs
-  seed: 42
-  output_dir: "runs"
-  compile: true
-
-# Model configuration
-model:
-  _target_: physicsnemo.models.transolver.Transolver
-  functional_dim: 2 # Input feature dimension
-  out_dim: 5        # Output feature dimension
-  embedding_dim: 3  # Spatial embedding dimension
-  n_layers: 8       # Number of transformer layers
-  n_hidden: 256     # Hidden dimension
-  dropout: 0.0      # Dropout rate
-  n_head: 8         # Number of attention heads
-  act: "gelu"       # Activation function
-  mlp_ratio: 4      # MLP ratio in attention blocks
-  slice_num: 128     # Number of slices in physics attention
-  unified_pos: false # Whether to use unified positional embeddings
-  ref: 8            # Reference dimension for unified pos
-  structured_shape: null
-  use_te: false      # Use transformer engine
-  Time_Input: false # Whether to use time embeddings
-
-scheduler:
-  name: "OneCycleLR"
-  params:
-    final_div_factor: 1e4
-
-# scheduler:
-#   name: "ReduceLROnPlateau"
-#   params:
-#     min_lr: 1e-6
-
-
-# Optimizer configuration
-optimizer:
-  _target_: torch.optim.AdamW
-  lr: 1.0e-3
-  weight_decay: 1.0e-4
-  betas: [0.9, 0.999]
-  eps: 1.0e-8
-
-# Data configuration
-data:
-  train:
-    data_path: /user_data/datasets/domino/train/
-  val:
-    data_path: /user_data/datasets/domino/val/
-  max_workers: 32
-  pin_memory: true
-  resolution: 200_000
-  mode: volume
-  data_keys:
-    - "volume_mesh_centers"
-    - "volume_fields"
-    - "air_density"
-    - "stream_velocity"
-  large_keys: 
-    - "volume_fields"
-    - "volume_mesh_centers"
-
-# Logging configuration
-logging:
-  level: INFO
-  format: '%(asctime)s - %(name)s - %(levelname)s - %(message)s'
-

From 0313c338429c649b2836f234d744705e352b01b5 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Thu, 31 Jul 2025 12:24:27 -0700
Subject: [PATCH 20/23] Bulk update of transolver model and example for CFD
 external aero.

---
 .../transolver/README.md                      | 208 +++++++-----
 .../transolver/conf/train_surface.yaml        |  37 +--
 .../transolver/datapipe.py                    |   1 -
 .../transolver/inference_on_stl.py            | 305 ++++++++++++++++++
 .../transolver/inference_on_zarr.py           | 169 ++++++++++
 .../external_aerodynamics/transolver/loss.py  |   2 +-
 .../transolver/metrics.py                     |  15 +-
 .../transolver/preprocess.py                  |  31 +-
 .../transolver/requirements.txt               |   4 +-
 .../external_aerodynamics/transolver/train.py | 159 ++++-----
 physicsnemo/models/transolver/transolver.py   |   2 +-
 11 files changed, 728 insertions(+), 205 deletions(-)
 create mode 100644 examples/cfd/external_aerodynamics/transolver/inference_on_stl.py
 create mode 100644 examples/cfd/external_aerodynamics/transolver/inference_on_zarr.py

diff --git a/examples/cfd/external_aerodynamics/transolver/README.md b/examples/cfd/external_aerodynamics/transolver/README.md
index 80b6a00474..40dfef7a0d 100644
--- a/examples/cfd/external_aerodynamics/transolver/README.md
+++ b/examples/cfd/external_aerodynamics/transolver/README.md
@@ -1,4 +1,6 @@
-# Transolver CFD Example: Code Overview
+# Transolver for External Aerodynamics on Irregular Meshes
+
+## Transolver CFD Example: Overview
 
 This directory contains the essential components for training and evaluating a
 Transolver model tailored to external aerodynamics CFD problems. The Transolver model
@@ -18,31 +20,115 @@ TransformerEngine.
 The training workflow for Transolver leverages the same input datasets as DoMINO. For
 more information on the datasets, refer to the external_aerodynamics example for DoMINO
 and the [PhysicsNeMo Curator](https://github.com/NVIDIA/physicsnemo-curator) for data
-preparation guidance. Unlike DoMINO, Transolver does not require neighbor points or any
-explicit structure (such as graph connections or edges) in the input data, resulting in
-a much simpler datapipe that is fully encapsulated within these examples.
-
-Below, we provide a high-level overview of the main files and their roles in the
-workflow.
-
----
-
-## 1. `src/train.py`
-
-This script serves as the main entry point for training the DoMINO/Transolver model,
-utilizing distributed data parallelism. The training loop processes the entire dataset,
-computing a simple point-wise relative L2 loss for either surface or volume data, and
-includes downsampling to handle the high native mesh resolutions.
-
-The script is designed for multi-GPU training using PyTorch’s DistributedDataParallel,
-and it loads datasets through a custom datapipe that supports distributed sampling.
-Model instantiation is flexible, supporting either surface or volume predictions
-as specified in the configuration.
-
-The script supports mixed-precision training with gradient
-scaling, and it manages checkpointing with the physicsnemo checkpointing utils.
-Throughout training, metrics are logged to both TensorBoard and the console and
-validation is performed after each epoch.
+preparation guidance.
+
+> **Note:** Currently, training transolver in this example supports **surface** data only.
+> Volumetric data is still in development.
+
+## Requirements
+
+Transolver requires TransformerEngine from NVIDIA, as well as Zarr >= 3.0 and `zarrs`
+for the data pipeline.  Install them with `pip install -r requirements.txt`
+
+> For the Transolver datapipe, zarr > 3.0 is required.  If you are using an older
+> container, you may need to `unset PIP_CONSTRAINTS` to allow zarr 3.0 or higher.
+
+## Using Transolver for External Aerodynamics
+
+1. Prepare the Dataset.  Follow the guidance from DoMINO and PhysicsNemo-Curator
+to prepare your dataset into a Zarr format.  We also use the dataset to compute
+normalization factors for the dataset, which help stabilize training.
+
+2. Train your model.  The model and training configuration is set in
+`conf/train_surface.yaml`, where you can control both network properties
+and training properties. See below for an overview and explanation of key
+parameters that may be of special interest.
+
+3. Use the trained model to perform inference.  This example contains two
+inference examples: one for inference on the validation set, already in
+Zarr format, and a second example for inference directly on .vtp files.
+
+The following sections contain further details on the training and inference
+recipe.
+
+## Model Training
+
+To train the model, first we compute normalization factors on the dataset to
+make the predictive quantities output in a well defined range.  The includec
+script, `compute_normalizations.py`, will compute either the normalization
+factors.  Once run, it should save to an output file similar to
+"surface_fields_normalization.npz".  This will get loaded during training.
+Check the training script to ensure the right path is used for your normalization
+factors - it's not a configuration parameter but directly encoded in the script.
+
+> By default, the normalization sets the mean to 0.0 and std to 1.0 of all labels
+> in the dataset, computing the mean across the train dataset.  You could adapt
+> this to a different normalization, however take care to update both the
+> preprocessing as well as inference scripts.  Min/Max is another popular strategy.
+
+To configure your training run, use `hydra` and `conf/train_surface.yaml`.  The
+config contains sections for the model, data, optimizer, and training settings.
+For details on the model parameters, see the API for `physicsnemo.models.transolver`.
+The data is processed with a custom Zarr dataloader, designed to use zarr 3.0 and
+`zarrs` rust implementation for an optimized Codec.  It also uses python's `threading`
+module to open parallel reads of multiple zarr keys.  You can control the number
+of parallel python threads via `data.max_workers`.
+
+Additionally, the Zarr dataloader optimizes CPU->GPU transfers by directly
+allocating pinned memory on the CPU, reading the Zarr data into that
+memory buffer via a 0-copy to numpy, and moving the data to GPU via a separate
+stream with non-blocking transfers.  In short: you can completely overlap IO
+and GPU processing as long as the IO file system can provide the next data example
+fast enough.  In reality, the IO latency has some variance but is not a bottleneck.
+
+You can disable memory pinning with `data.pin_memory=False`.  Further, to fit
+the training into memory, you can apply on-the-fly downsampling to the data
+with `data.resolution=N`, where `N` is how many points per GPU to use.  This dataloader
+will yield the full data examples in shapes of `[1, K, f]` where `K` is the resolution
+of the mesh, and `f` is the feature space (3 for points, normals, etc.  4 for surface
+fields).  Downsampling happens in the preprocessing pipeline.
+
+> The pipeline has the ability to optimally load data from disk into `physicsnemo.ShardTensor`
+> for domain parallelism - however the model support is still in development.
+
+During training, the configuration uses the OneCyle learning rate (similar to the
+original Transolver publication), and float32 format.  The scheduler and learning rate
+may be configured - note that the scheduler is updated every training step.  For
+schedulers that update every epoch, modification of the training script may be required.
+
+### Training Precision
+
+Transolver, as a transformer-like architecture, has support for NVIDIA's
+[TransformerEngine](https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/index.html)
+built in.  You can enable/disable the transformer engine path in the model with
+`model.use_te=[True | False]`.  Available precisions for training with `transformer_engine`
+are `training.precision=["float32" | "float16" | "bfloat16" | "float8" ]`.  In `float8`
+precision, the TransformerEngine Hybrid recipe is used for casting weights and inputs
+in the forward and backwards passes.  For more details on `float8` precision, see
+the fp8 guide from
+[TransformerEngine](https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/examples/fp8_primer.html).
+When using fp8, the training script will automatically pad and unpad the input and output,
+respectively, to use the fp8 hardware correctly.
+
+> **Float8** precisions are only available on GPUs with fp8 tensorcore support, such
+> as Hopper, Blackwell, Ada Lovelace, and others.
+
+### Other Configuration Settings
+
+Several other important configuration settings are available:
+
+- `training.compile` will use `torch.compile` for optimized performance.  It is not
+compatible with `transformer_engine` (`model.use_te=True`).  If TransformerEngine is
+not used, and half precision is, `torch.compile` is recommended for improved performance.
+- `training.num_epochs` controls the total number of epochs used during training.
+- `training.save_interval` will dictate how often the model weights and training
+tools are checkpointed.
+
+> **Note** Like other parameters of the model, changing the value of `model.use_te`
+> will make checkpoints incompatible.
+
+The training script supports data-parallel training via PyTorch DDP.  In a future
+update, we may enable domain parallelism via FSDP and ShardTensor.
 
 The script can be launched on a single GPU with
 
@@ -97,58 +183,28 @@ Epoch 47 Validation Average Metrics:
 +-------------+---------------------+
 ```
 
----
-
-## 2. `conf/train_volume.yaml` and `conf/train_surface.yaml`
-
-These configuration files define all the settings required for a training run. They
-specify output directories, run identifiers, random seeds, and precision settings, as
-well as the number of epochs, checkpoint intervals, and whether to use compilation. The
-model architecture is described here, including input and output dimensions, the number
-of layers, embedding size, attention heads, and activation functions. Optimizer
-settings such as the type (AdamW), learning rate, weight decay, and other
-hyperparameters are also included. Data-related settings cover paths to the training and
-validation datasets, worker and thread counts, memory pinning, and which data keys to
-load. Finally, logging preferences are set, controlling the level and format of output.
-
-Note that, to use TransformerEngine, you must enable it directly in the model settings.
-TransformerEngine is incompatible with `torch.compile`.  
-
----
-
-## 3. `loss.py`
-
-This file defines the loss functions used during Transolver training, primarily
-focusing on a relative L2 loss. For surface data, it computes mean squared error (MSE)
-or root mean squared error (RMSE) for both pressure (a scalar) and wall shear (a
-vector), handling each component separately before combining the results.
-
----
-
-## 4. `metrics.py`
+## Dataset Inference
 
-Evaluation metrics for model predictions are defined here. The metrics are designed to
-aggregate results across distributed processes, ensuring consistency in multi-GPU
-setups. For surface predictions, the script computes normalized L2 errors for both
-pressure and shear components, after unnormalizing the predictions and targets. The
-structure of the code allows for straightforward extension to additional metrics or
-application domains as needed.
+There are two scripts provided as inference examples - it's expected that every user's
+inference workloads are different, so these aim to cover common scenarios as examples.
 
----
+First, the validation dataset in Zarr format can be loaded, processed, and the L2
+metrics summarized in `inference_on_zarr.py`.  Alternatively, the model can be used
+directly on `.vtp` or `.stl` files as shown in `inference_on_vtp.py`.  Note that the
+script contains several parameters from the DrivaerML dataset as hardcoded variable
+names: `CpMeanTrim`, `pMeanTrim`, `wallShearStressMeanTrim`, which are used to
+compute the L2 metrics on the inference outputs.
 
-## 5. `datapipe.py`
+In `inference_on_zarr.py`, the dataset examples are downsampled and preprocessed
+exactly as in the training script.  In `inference_on_stl.py`, however, the entire
+mesh is processed.  To enable the mesh to fit into GPU memory, the mesh is chunked
+into pieces that are then processed, and recombined to form the prediction on the
+entire mesh.  The outputs are then saved to .vtp files for downstream analysis.
 
-Efficient loading of large CFD datasets stored in Zarr format is handled by this file,
-which implements like a PyTorch dataset. The datapipe reads large arrays in a chunk-aligned
-manner for optimal performance, leveraging threads for parallel I/O. It offers
-flexibility in specifying which data keys to load and which are considered “large” (and
-thus read in chunks, vs. small arrays which are directly read in their entirety).
-For faster GPU transfers, it can allocate directly to pinned memory, then share buffers
-with numpy for streaming from disk directly to pinned memory.  The dataload has a
-prefetch utility, as well, enabling the dataloader to queue the next batch to GPU.
+## Future work
 
-CPU to GPU transfers are performed in a separate CUDA stream from the main computation,
-enabling async transfers overlapping with model training.  The main parameter to tune
-is the number of workers for the threading: too many, and you will introduce
-CPU overhead limiting the model performance.  Too few, and the dataload won't acheive
-peak throughput of dataloading.
+The Transolver model is a promising, transformer-based model that produces high
+quality predictions for CFD surrogate simulations.  In the future, we may update
+the example to include domain parallelism and Transolver++ enhancements,
+as well as volumetric data examples.  If you
+have issues, requests, or other items please feel free to open an issue and discuss!
diff --git a/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
index 76b9c6e691..67d751fc50 100644
--- a/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
+++ b/examples/cfd/external_aerodynamics/transolver/conf/train_surface.yaml
@@ -17,16 +17,14 @@
 
 
 output_dir: "runs"
-run_id: "surface/ReduceLR/TE/float16-3/"
+run_id: "surface/precision-convergence/float32"
 
 # Training configuration
 training:
-  precision: float16 # float32, float16, bfloat16
-  num_epochs: 500
+  precision: float32 # float32, float16, bfloat16
+  num_epochs: 251 # Add one to save at 250 :)
   save_interval: 25  # Save checkpoint every N epochs
-  seed: 42
-  output_dir: "runs"
-  compile: false
+  compile: True
 
 # Model configuration
 model:
@@ -44,26 +42,26 @@ model:
   unified_pos: false # Whether to use unified positional embeddings
   ref: 8            # Reference dimension for unified pos
   structured_shape: null
-  use_te: True      # Use transformer engine
+  use_te: false      # Use transformer engine
   Time_Input: false # Whether to use time embeddings
 
-# scheduler:
-#   name: "OneCycleLR"
-#   params:
-#     final_div_factor: 1e4
-
-# StepLR scheduler: Decays the learning rate by gamma every step_size epochs
 scheduler:
-  name: "StepLR"
+  name: "OneCycleLR"
   params:
-    step_size: 50     # Decay every 100 epochs (set X as desired)
-    gamma: 0.5        # Decay factor
+    final_div_factor: 1e4
+
+# # StepLR scheduler: Decays the learning rate by gamma every step_size epochs
+# scheduler:
+#   name: "StepLR"
+#   params:
+#     step_size: 50     # Decay every 100 epochs (set X as desired)
+#     gamma: 0.5        # Decay factor
 
 
 # Optimizer configuration
 optimizer:
   _target_: torch.optim.AdamW
-  lr: 3.0e-4
+  lr: 5.0e-4
   weight_decay: 1.0e-4
   betas: [0.9, 0.999]
   eps: 1.0e-8
@@ -74,10 +72,9 @@ data:
     data_path: /user_data/datasets/domino/train/
   val:
     data_path: /user_data/datasets/domino/val/
-  max_workers: 64
+  max_workers: 8
   pin_memory: true
-  # resolution: 10000
-  resolution: 325_000
+  resolution: 400_000
   mode: surface
   data_keys:
     - "surface_fields"
diff --git a/examples/cfd/external_aerodynamics/transolver/datapipe.py b/examples/cfd/external_aerodynamics/transolver/datapipe.py
index b1b65a1c3a..36772f961a 100644
--- a/examples/cfd/external_aerodynamics/transolver/datapipe.py
+++ b/examples/cfd/external_aerodynamics/transolver/datapipe.py
@@ -660,7 +660,6 @@ def __getitem__(self, idx: int) -> Dict[str, torch.Tensor | ShardTensor]:
         else:
             filename = self.filenames[idx]
             filepath = self.data_path / filename
-
             # Read data from zarr file
             data = self._read_zarr_file(filepath)
             data = self._move_to_gpu(data)
diff --git a/examples/cfd/external_aerodynamics/transolver/inference_on_stl.py b/examples/cfd/external_aerodynamics/transolver/inference_on_stl.py
new file mode 100644
index 0000000000..b776a9ef22
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/inference_on_stl.py
@@ -0,0 +1,305 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+import torch
+
+import hydra
+from omegaconf import DictConfig
+
+import pyvista as pv
+from physicsnemo.models.transolver.transolver import Transolver
+from physicsnemo.launch.utils import load_checkpoint
+
+from physicsnemo.distributed import DistributedManager
+
+import vtk
+from vtk.util import numpy_support
+import os
+import math
+import time
+
+
+def read_data_from_stl(stl_path, air_density=1.2050, stream_velocity=30.0):
+
+    dm = DistributedManager()
+
+    mesh = pv.read(stl_path)
+
+    batch = {}
+
+    batch["surface_mesh_centers"] = np.asarray(mesh.cell_centers().points)
+
+    normals = np.asarray(mesh.cell_normals)
+    # Normalize cell normals
+    surface_normals = (
+        surface_normals / np.linalg.norm(surface_normals, axis=1)[:, np.newaxis]
+    )
+    batch["surface_normals"] = surface_normals
+    surface_areas = mesh.compute_cell_sizes(length=False, area=True, volume=False)
+    batch["surface_mesh_sizes"] = np.array(surface_areas.cell_data["Area"])
+
+    batch["air_density"] = np.array([air_density], dtype="float32")
+    batch["stream_velocity"] = np.array([stream_velocity], dtype="float32")
+
+    batch = {
+        k: torch.from_numpy(v).to(device=dm.device, dtype=torch.float32)
+        for k, v in batch.items()
+    }
+
+    batch = {k: torch.unsqueeze(v, dim=0) for k, v in batch.items()}
+
+    return batch
+
+
+def read_data_from_vtp(vtp_path, air_density=1.2050, stream_velocity=30.0):
+
+    dm = DistributedManager()
+
+    mesh = pv.read(vtp_path)
+
+    batch = {}
+
+    batch["surface_mesh_centers"] = np.asarray(mesh.cell_centers().points)
+    batch["surface_normals"] = np.asarray(mesh.cell_normals)
+    surface_areas = mesh.compute_cell_sizes(length=False, area=True, volume=False)
+    batch["surface_mesh_sizes"] = np.array(surface_areas.cell_data["Area"])
+
+    batch["CpMeanTrim"] = np.asarray(mesh.cell_data["CpMeanTrim"])
+    batch["pMeanTrim"] = np.asarray(mesh.cell_data["pMeanTrim"])
+    batch["wallShearStressMeanTrim"] = np.asarray(
+        mesh.cell_data["wallShearStressMeanTrim"]
+    )
+
+    batch["air_density"] = np.array([air_density], dtype="float32")
+    batch["stream_velocity"] = np.array([stream_velocity], dtype="float32")
+
+    # From VTP we can also exctract the ground-truth results:
+    batch = {
+        k: torch.from_numpy(v).to(device=dm.device, dtype=torch.float32)
+        for k, v in batch.items()
+    }
+
+    batch = {k: torch.unsqueeze(v, dim=0) for k, v in batch.items()}
+
+    return mesh, batch
+
+
+def preprocess_data(
+    batch,
+):
+
+    mesh_centers = batch["surface_mesh_centers"]
+    normals = batch["surface_normals"]
+    node_features = torch.stack(
+        [batch["air_density"], batch["stream_velocity"]], axis=-1
+    )
+
+    # Calculate center of mass
+    sizes = batch["surface_mesh_sizes"]
+
+    total_weighted_position = torch.einsum("ki,kij->kj", sizes, mesh_centers)
+    total_size = torch.sum(sizes)
+    center_of_mass = total_weighted_position[None, ...] / total_size
+
+    # Subtract the COM from the centers:
+    mesh_centers = mesh_centers - center_of_mass
+
+    embeddings = torch.cat(
+        [
+            mesh_centers,
+            normals,
+        ],
+        dim=-1,
+    )
+    node_features = node_features.expand(1, embeddings.shape[1], -1)
+
+    return node_features, embeddings
+
+
+def process_vtp_file(
+    vtp_file, model, norm_factors, dist_manager, output_folder, batch_size=500_000
+):
+
+    # First, load the data and mesh from the file:
+    try:
+        mesh, batch = read_data_from_vtp(vtp_file)
+    except FileNotFoundError as e:
+        print(f"File not found: {vtp_file}")
+        return
+
+    # Run preprocessing to prepare the data for the model:
+    fx, embedding = preprocess_data(batch)
+
+    with torch.no_grad():
+
+        if batch_size > fx.shape[1]:
+
+            prediction = (
+                model(fx, embedding) * norm_factors["std"] + norm_factors["mean"]
+            )
+
+        else:
+            # Split the indices by a batch size.  We shuffle the cells into
+            # the batches (don't forget to unshuffle later!)
+            indices = torch.randperm(fx.shape[1], device=fx.device)
+
+            index_blocks = torch.split(indices, batch_size)
+
+            predictions = []
+            for i, index_block in enumerate(index_blocks):
+                local_fx = fx[:, index_block]
+                local_embedding = embedding[:, index_block]
+                predictions.append(
+                    model(local_fx, local_embedding) * norm_factors["std"]
+                    + norm_factors["mean"]
+                )
+
+            prediction = torch.cat(predictions, dim=1)
+
+            # Now, we have to *unshuffle* the prediction to the original index
+            inverse_indices = torch.empty_like(indices)
+            inverse_indices[indices] = torch.arange(
+                indices.size(0), device=indices.device
+            )
+            # Suppose prediction is of shape [batch, N, ...]
+            prediction = prediction[:, inverse_indices]
+
+        pred_pressure, pred_shear = torch.split(prediction, (1, 3), dim=-1)
+        pred_pressure = pred_pressure.squeeze(-1)
+
+        pred_pressure = pred_pressure * (
+            batch["air_density"] * batch["stream_velocity"] ** 2
+        )
+        pred_shear = pred_shear * (batch["air_density"] * batch["stream_velocity"] ** 2)
+        target_pressure = batch["pMeanTrim"]
+        target_shear = batch["wallShearStressMeanTrim"]
+
+        pressure_l2_num = (pred_pressure - target_pressure) ** 2
+        pressure_l2_num = torch.sum(pressure_l2_num, dim=1)
+        pressure_l2_num = torch.sqrt(pressure_l2_num)
+
+        pressure_l2_denom = target_pressure**2
+        pressure_l2_denom = torch.sum(pressure_l2_denom, dim=1)
+        pressure_l2_denom = torch.sqrt(pressure_l2_denom)
+
+        pressure_l2 = pressure_l2_num / pressure_l2_denom
+
+        shear_l2_num = (pred_shear - target_shear) ** 2
+        shear_l2_num = torch.sum(shear_l2_num, dim=1)
+        shear_l2_num = torch.sqrt(shear_l2_num)
+
+        shear_l2_denom = target_shear**2
+        shear_l2_denom = torch.sum(shear_l2_denom, dim=1)
+        shear_l2_denom = torch.sqrt(shear_l2_denom)
+
+        shear_l2 = shear_l2_num / shear_l2_denom
+
+        print(f"pressure l2: {pressure_l2}")
+        print(f"shear l2: {shear_l2}")
+
+    # Write the output to a new .vtp file.  Clone the old information:
+    output_mesh = mesh.copy()
+    # Convert tensors to numpy arrays and squeeze batch dimension
+    pred_pressure_np = pred_pressure[0].cpu().numpy()
+    pred_shear_np = pred_shear[0].cpu().numpy()
+    # Add arrays to the mesh as cell data
+    output_mesh.cell_data["PredictedPressure"] = pred_pressure_np
+    output_mesh.cell_data["PredictedShear"] = pred_shear_np
+    # Ensure the output directory exists
+    os.makedirs(output_folder, exist_ok=True)
+    # Construct output file path
+    base_name = os.path.basename(vtp_file)
+    output_path = os.path.join(output_folder, f"pred_{base_name}")
+    # Write to file
+    output_mesh.save(output_path)
+    print(f"Saved prediction VTP to: {output_path}")
+
+
+def inference_on_vtp_or_stl(cfg: DictConfig):
+
+    DistributedManager.initialize()
+
+    dist_manager = DistributedManager()
+
+    run_id = cfg.run_id
+
+    # Set up model
+    model = hydra.utils.instantiate(cfg.model)
+    model.eval()
+    model.to(dist_manager.device)
+
+    ckpt_args = {
+        "path": f"{cfg.output_dir}/{cfg.run_id}/checkpoints",
+        "models": model,
+    }
+
+    # Load the normalization factors:
+    norm_file = "surface_fields_normalization.npz"
+    norm_data = np.load(norm_file)
+    norm_factors = {
+        "mean": torch.from_numpy(norm_data["mean"]).to(dist_manager.device),
+        "std": torch.from_numpy(norm_data["std"]).to(dist_manager.device),
+    }
+
+    # Restore the model:
+    loaded_epoch = load_checkpoint(device=dist_manager.device, **ckpt_args)
+    print(f"loaded epoch: {loaded_epoch}")
+
+    stl_input_path = "/group_data/datasets/drivaer_aws/drivaer_data_full/"
+    vtp_output_path = f"/user_data/datasets/drivaer_aws/drivaer_data_full/{run_id}/"
+
+    all_files = list(range(1, 501))
+
+    all_files = [stl_input_path + f"/run_{i}/boundary_{i}.vtp" for i in all_files]
+
+    # Remove files that already exist in the output directory
+    filtered_files = []
+    for file_path in all_files:
+        base_name = os.path.basename(file_path)
+        output_path = os.path.join(vtp_output_path, f"pred_{base_name}")
+        if not os.path.exists(output_path):
+            filtered_files.append(file_path)
+    all_files = filtered_files
+
+    # print(f"all files: {all_files} of length {len(all_files)}")
+
+    this_device_files = all_files[dist_manager.rank :: dist_manager.world_size]
+
+    print(
+        f"Rank {dist_manager.rank} of {dist_manager.world_size} is processing {len(this_device_files)} files"
+    )
+    for vtp_file in this_device_files:
+        start = time.time()
+
+        # Process files:
+        process_vtp_file(vtp_file, model, norm_factors, dist_manager, vtp_output_path)
+        end = time.time()
+        print(f"time taken: {end - start}")
+
+
+@hydra.main(version_base=None, config_path="conf", config_name="train_surface")
+def launch(cfg: DictConfig):
+    """Launch training with hydra configuration
+
+    Args:
+        cfg: Hydra configuration object
+    """
+    inference_on_vtp_or_stl(cfg)
+
+
+if __name__ == "__main__":
+    launch()
diff --git a/examples/cfd/external_aerodynamics/transolver/inference_on_zarr.py b/examples/cfd/external_aerodynamics/transolver/inference_on_zarr.py
new file mode 100644
index 0000000000..2e9549f10f
--- /dev/null
+++ b/examples/cfd/external_aerodynamics/transolver/inference_on_zarr.py
@@ -0,0 +1,169 @@
+# SPDX-FileCopyrightText: Copyright (c) 2023 - 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+import torch
+
+import hydra
+from omegaconf import DictConfig
+from physicsnemo.models.transolver.transolver import Transolver
+from physicsnemo.launch.utils import load_checkpoint
+
+from physicsnemo.distributed import DistributedManager
+
+import time
+
+from datapipe import DomainParallelZarrDataset
+
+from train import forward_pass
+from tabulate import tabulate
+
+
+def inference(cfg: DictConfig) -> None:
+    """
+    Run inference on a validation Zarr dataset using a trained Transolver model.
+
+    Args:
+        cfg (DictConfig): Hydra configuration object containing model, data, and training settings.
+
+    Returns:
+        None
+    """
+    DistributedManager.initialize()
+
+    dist_manager = DistributedManager()
+
+    # Set up model
+    model = hydra.utils.instantiate(cfg.model)
+    model.eval()
+    model.to(dist_manager.device)
+
+    # Validation dataset
+
+    val_dataset = DomainParallelZarrDataset(
+        data_path=cfg.data.val.data_path,  # Assuming validation data path is configured
+        device_mesh=None,
+        placements=None,
+        max_workers=cfg.data.max_workers,
+        pin_memory=cfg.data.pin_memory,
+        keys_to_read=cfg.data.data_keys,
+        large_keys=cfg.data.large_keys,
+    )
+
+    ckpt_args = {
+        "path": f"{cfg.output_dir}/{cfg.run_id}/checkpoints",
+        "models": model,
+    }
+
+    # Load the normalization factors:
+    norm_file = "surface_fields_normalization.npz"
+    norm_data = np.load(norm_file)
+    norm_factors = {
+        "mean": torch.from_numpy(norm_data["mean"]).to(dist_manager.device),
+        "std": torch.from_numpy(norm_data["std"]).to(dist_manager.device),
+    }
+
+    loaded_epoch = load_checkpoint(device=dist_manager.device, **ckpt_args)
+    print(f"loaded epoch: {loaded_epoch}")
+
+    results = []
+    start = time.time()
+    for batch_idx in range(len(val_dataset)):
+
+        batch = val_dataset[batch_idx]
+        with torch.no_grad():
+            loss, metrics = forward_pass(
+                batch,
+                model,
+                cfg.training.precision,
+                None,
+                dist_manager,
+                cfg,
+                norm_factors,
+            )
+
+        # Extract metric values and convert tensors to floats
+        l2_pressure = (
+            metrics["l2_pressure"].item()
+            if hasattr(metrics["l2_pressure"], "item")
+            else metrics["l2_pressure"]
+        )
+        l2_shear_x = (
+            metrics["l2_shear_x"].item()
+            if hasattr(metrics["l2_shear_x"], "item")
+            else metrics["l2_shear_x"]
+        )
+        l2_shear_y = (
+            metrics["l2_shear_y"].item()
+            if hasattr(metrics["l2_shear_y"], "item")
+            else metrics["l2_shear_y"]
+        )
+        l2_shear_z = (
+            metrics["l2_shear_z"].item()
+            if hasattr(metrics["l2_shear_z"], "item")
+            else metrics["l2_shear_z"]
+        )
+
+        end = time.time()
+        elapsed = end - start
+        print(f"Finished batch {batch_idx} in {elapsed:.4f} seconds")
+        results.append(
+            [
+                batch_idx,
+                f"{loss:.4f}",
+                f"{l2_pressure:.4f}",
+                f"{l2_shear_x:.4f}",
+                f"{l2_shear_y:.4f}",
+                f"{l2_shear_z:.4f}",
+                f"{elapsed:.4f}",
+            ]
+        )
+
+        start = time.time()
+
+    headers = [
+        "Batch",
+        "Loss",
+        "L2 Pressure",
+        "L2 Shear X",
+        "L2 Shear Y",
+        "L2 Shear Z",
+        "Elapsed (s)",
+    ]
+    print(tabulate(results, headers=headers, tablefmt="github"))
+
+    # Calculate means for each metric (skip batch index)
+    if results:
+        # Convert string values back to float for mean calculation
+        arr = np.array(results)[:, 1:].astype(float)
+        means = arr.mean(axis=0)
+        mean_row = ["Mean"] + [f"{m:.4f}" for m in means]
+        print(tabulate([mean_row], headers=headers, tablefmt="github"))
+
+
+@hydra.main(version_base=None, config_path="conf", config_name="train_surface")
+def launch(cfg: DictConfig) -> None:
+    """
+    Launch inference with Hydra configuration.
+
+    Args:
+        cfg (DictConfig): Hydra configuration object.
+    """
+    inference(cfg)
+
+
+if __name__ == "__main__":
+    launch()
diff --git a/examples/cfd/external_aerodynamics/transolver/loss.py b/examples/cfd/external_aerodynamics/transolver/loss.py
index 37dd31d969..f81632f4da 100644
--- a/examples/cfd/external_aerodynamics/transolver/loss.py
+++ b/examples/cfd/external_aerodynamics/transolver/loss.py
@@ -38,7 +38,7 @@ def loss_fn(
     if mode == "surface":
         loss = loss_fn_surface(pred, target, "mse")
     elif mode == "volume":
-        loss = loss_fn_volume(pred, target, "rmse")
+        loss = loss_fn_volume(pred, target, "mse")
     # 100 * integral_loss_fn(pred, target, others["surface_areas"], others["surface_normals"], others["stream_velocity"])
     return loss
 
diff --git a/examples/cfd/external_aerodynamics/transolver/metrics.py b/examples/cfd/external_aerodynamics/transolver/metrics.py
index 5275710f06..219d315825 100644
--- a/examples/cfd/external_aerodynamics/transolver/metrics.py
+++ b/examples/cfd/external_aerodynamics/transolver/metrics.py
@@ -60,6 +60,7 @@ def metrics_fn(
     others: dict[str, torch.Tensor],
     dm: DistributedManager,
     mode: str,
+    norm_factors: dict[str, torch.Tensor],
 ) -> dict[str, torch.Tensor]:
     """
     Computes metrics for either surface or volume data.
@@ -76,9 +77,9 @@ def metrics_fn(
     """
     with torch.no_grad():
         if mode == "surface":
-            metrics = metrics_fn_surface(pred, target, others, dm)
+            metrics = metrics_fn_surface(pred, target, others, dm, norm_factors)
         elif mode == "volume":
-            metrics = metrics_fn_volume(pred, target, others, dm)
+            metrics = metrics_fn_volume(pred, target, others, dm, norm_factors)
         else:
             raise ValueError(f"Unknown data mode: {mode}")
 
@@ -91,6 +92,7 @@ def metrics_fn_volume(
     target: torch.Tensor,
     others: dict[str, torch.Tensor],
     dm: DistributedManager,
+    norm_factors: dict[str, torch.Tensor],
 ) -> dict[str, torch.Tensor]:
     """
     Computes L2 volume metrics between prediction and target.
@@ -104,8 +106,8 @@ def metrics_fn_volume(
     Returns:
         Dictionary of L2 volume metrics.
     """
-    target = target * others["norm_std"] + others["norm_mean"]
-    pred = pred * others["norm_std"] + others["norm_mean"]
+    target = target * norm_factors["std"] + norm_factors["mean"]
+    pred = pred * norm_factors["std"] + norm_factors["mean"]
 
     l2_num = (pred - target) ** 2
     l2_num = torch.sum(l2_num, dim=1)
@@ -133,6 +135,7 @@ def metrics_fn_surface(
     target: torch.Tensor,
     others: dict[str, torch.Tensor],
     dm: DistributedManager,
+    norm_factors: dict[str, torch.Tensor],
 ) -> dict[str, torch.Tensor]:
     """
     Computes L2 surface metrics between prediction and target.
@@ -147,8 +150,8 @@ def metrics_fn_surface(
         Dictionary of L2 surface metrics.
     """
     # Unnormalize the surface values for L2:
-    target = target * others["norm_std"] + others["norm_mean"]
-    pred = pred * others["norm_std"] + others["norm_mean"]
+    target = target * norm_factors["std"] + norm_factors["mean"]
+    pred = pred * norm_factors["std"] + norm_factors["mean"]
 
     l2_num = (pred - target) ** 2
     l2_num = torch.sum(l2_num, dim=1)
diff --git a/examples/cfd/external_aerodynamics/transolver/preprocess.py b/examples/cfd/external_aerodynamics/transolver/preprocess.py
index 0121807e06..5c6185f915 100644
--- a/examples/cfd/external_aerodynamics/transolver/preprocess.py
+++ b/examples/cfd/external_aerodynamics/transolver/preprocess.py
@@ -23,6 +23,7 @@
 @profile
 def preprocess_surface_data(
     batch: dict,
+    norm_factors: dict[str, torch.Tensor],
 ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
 
     """
@@ -41,9 +42,7 @@ def preprocess_surface_data(
     ).to(torch.float32)
 
     # Normalize the surface fields:
-    norm_mean = targets.mean(dim=1)
-    norm_std = torch.sqrt(torch.mean((targets - norm_mean.unsqueeze(1)) ** 2, dim=1))
-    targets = (targets - norm_mean) / norm_std
+    targets = (targets - norm_factors["mean"]) / norm_factors["std"]
 
     # If you want to use this, be sure to updat the
     # functional_dim value in your configuration
@@ -84,21 +83,14 @@ def preprocess_surface_data(
         "surface_normals": normals,
         "stream_velocity": batch["stream_velocity"],
         "air_density": batch["air_density"],
-        "norm_mean": norm_mean,
-        "norm_std": norm_std,
     }
-    print(
-        f"node_features shape: {node_features.shape} nd placements: {node_features._spec.placements}"
-    )
-    print(
-        f"embeddings shape: {embeddings.shape} nd placements: {embeddings._spec.placements}"
-    )
-    print(f"targets shape: {targets.shape} nd placements: {targets._spec.placements}")
+
     return node_features, embeddings, targets, others
 
 
 def preprocess_volume_data(
     batch: dict,
+    norm_factors: dict[str, torch.Tensor],
 ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, dict]:
     """
     Preprocess the volumetric data.  Right now, it's just
@@ -120,16 +112,10 @@ def preprocess_volume_data(
 
     embeddings = torch.cat([mesh_centers, sdf], dim=-1)
 
-    # Normalize the surface fields:
-    norm_mean = targets.mean(dim=1)
-    # norm_std = targets.std(dim=1)
-    norm_std = torch.sqrt(torch.mean((targets - norm_mean.unsqueeze(1)) ** 2, dim=1))
-
-    targets = (targets - norm_mean) / norm_std
+    # Normalize the volume fields:
+    targets = (targets - norm_factors["mean"]) / norm_factors["std"]
 
     others = {
-        "norm_mean": norm_mean,
-        "norm_std": norm_std,
         "stream_velocity": batch["stream_velocity"],
         "air_density": batch["air_density"],
     }
@@ -143,6 +129,11 @@ def downsample_surface(
     targets: torch.Tensor,
     num_keep=1024,
 ):
+
+    if num_keep == -1:
+        features = features.unsqueeze(1).expand(1, embeddings.shape[1], -1)
+        return features, embeddings, targets
+
     """
     Downsample the surface data. We generate one set of indices, and
     use it to sample the same points from the features, embeddings,
diff --git a/examples/cfd/external_aerodynamics/transolver/requirements.txt b/examples/cfd/external_aerodynamics/transolver/requirements.txt
index e0414bda88..3d3f33a9b1 100644
--- a/examples/cfd/external_aerodynamics/transolver/requirements.txt
+++ b/examples/cfd/external_aerodynamics/transolver/requirements.txt
@@ -3,4 +3,6 @@ tabulate
 tensorboard
 termcolor
 einops
-transformer_engine[pytorch]
\ No newline at end of file
+transformer_engine[pytorch]
+zarr>=3.0
+zarrs
\ No newline at end of file
diff --git a/examples/cfd/external_aerodynamics/transolver/train.py b/examples/cfd/external_aerodynamics/transolver/train.py
index 0a7702845a..42f1b8d48b 100644
--- a/examples/cfd/external_aerodynamics/transolver/train.py
+++ b/examples/cfd/external_aerodynamics/transolver/train.py
@@ -23,6 +23,7 @@
 from omegaconf import DictConfig
 from torch.utils.tensorboard import SummaryWriter
 
+import numpy as np
 
 from torch.distributed.fsdp import (
     FullyShardedDataParallel as FSDP,
@@ -34,7 +35,6 @@
 from physicsnemo.launch.utils import load_checkpoint, save_checkpoint
 from physicsnemo.launch.logging import PythonLogger, RankZeroLoggingWrapper
 from physicsnemo.distributed import DistributedManager
-from typing import Literal
 
 from physicsnemo.utils.profiling import profile, Profiler
 
@@ -108,45 +108,54 @@ def forward_pass(
     output_pad_size: int | None,
     dist_manager: DistributedManager,
     cfg: DictConfig,
+    norm_factors: dict[str, torch.Tensor],
 ):
     """
     Run the forward pass of the model for one batch, including metrics and loss calculation.
     """
 
     if cfg.data.mode == "surface":
-        features, embeddings, targets, others = preprocess_surface_data(batch)
+        features, embeddings, targets, others = preprocess_surface_data(
+            batch, norm_factors
+        )
         features, embeddings, targets = downsample_surface(
             features, embeddings, targets, cfg.data.resolution
         )
 
     elif cfg.data.mode == "volume":
-        features, embeddings, targets, others = preprocess_volume_data(batch)
+        features, embeddings, targets, others = preprocess_volume_data(
+            batch, norm_factors
+        )
         features, embeddings, targets = downsample_volume(
             features, embeddings, targets, cfg.data.resolution
         )
     else:
         raise ValueError(f"Unknown data mode: {cfg.data.mode}")
 
+    # del batch
+
     # Cast precisions:
     features, embeddings = cast_precisions(features, embeddings, precision)
     with get_autocast_context(precision):
-        print(
-            f"features shape: {features.shape} nd placements: {features._spec.placements}"
-        )
-        print(
-            f"embeddings shape: {embeddings.shape} nd placements: {embeddings._spec.placements}"
-        )
-        print(
-            f"targets shape: {targets.shape} nd placements: {targets._spec.placements}"
-        )
-        print(f"cat shape: {torch.cat((features, embeddings), dim=-1).shape}")
+
+        # For fp8, we may have to pad the inputs:
+        if precision == "float8":
+            fx_dim = features.shape[-1] + embeddings.shape[-1]
+            if fx_dim % 16 != 0:
+                pad_size = 16 - (fx_dim % 16)
+                features = torch.nn.functional.pad(features, (0, pad_size))
+                fx_dim = features.shape[-1] + embeddings.shape[-1]
+
         outputs = model(features, embeddings)
+
         if output_pad_size is not None:
             # Remove the padded outputs:
             outputs = outputs[:, :, :-output_pad_size]
         loss = loss_fn(outputs, targets, others, cfg.data.mode)
 
-    metrics = metrics_fn(outputs, targets, others, dist_manager, cfg.data.mode)
+    metrics = metrics_fn(
+        outputs, targets, others, dist_manager, cfg.data.mode, norm_factors
+    )
 
     return loss, metrics
 
@@ -154,7 +163,7 @@ def forward_pass(
 @profile
 def train_epoch(
     dataloader,
-    sampler: torch.utils.data.Sampler | None,
+    sampler: torch.utils.data.Sampler,
     model: torch.nn.Module,
     output_pad_size: int | None,
     optimizer: torch.optim.Optimizer,
@@ -164,6 +173,7 @@ def train_epoch(
     epoch: int,
     cfg: DictConfig,
     dist_manager: DistributedManager,
+    norm_factors: dict[str, torch.Tensor],
     scaler: GradScaler | None = None,
 ) -> float:
     """
@@ -171,7 +181,7 @@ def train_epoch(
 
     Args:
         dataloader (list[dict]): Training data loader
-        sampler (torch.utils.data.Sampler | None): Sampler for distributed or sequential sampling.
+        sampler (torch.utils.data.Sampler): Sampler for distributed or sequential sampling.
         model (torch.nn.Module): The neural network model to train.
         output_pad_size (int | None): Optional output padding size for lowest precisions (FP8).
         optimizer (torch.optim.Optimizer): Optimizer for model parameters.
@@ -181,8 +191,8 @@ def train_epoch(
         epoch (int): Current epoch number.
         cfg (DictConfig): Hydra configuration object.
         dist_manager (DistributedManager): Distributed manager from physicsnemo.
+        norm_factors (dict[str, torch.Tensor]): Normalization factors for the data.
         scaler (GradScaler | None, optional): Gradient scaler for mixed precision training.
-
     Returns:
         float: The average training loss for the epoch.
     """
@@ -196,16 +206,20 @@ def train_epoch(
     start_time = time.time()
     with Profiler():
         for i, batch_idx in enumerate(epoch_indices):
-
-            if i > 10:
-                break
             batch = dataloader[batch_idx]
+
             # preload the next batch, if we're not on the last batch
             if i < epoch_len - 1 and sampler is not None:
                 dataloader.preload(epoch_indices[i + 1])
 
             loss, metrics = forward_pass(
-                batch, model, precision, output_pad_size, dist_manager, cfg
+                batch,
+                model,
+                precision,
+                output_pad_size,
+                dist_manager,
+                cfg,
+                norm_factors,
             )
 
             optimizer.zero_grad()
@@ -237,8 +251,10 @@ def train_epoch(
             start_time = end_time
             images_per_second = 1 / duration
 
+            mem_usage = torch.cuda.memory_reserved() / 1024**3
+
             logger.info(
-                f"Epoch {epoch} [{i}/{epoch_len}] Loss: {this_loss:.6f} Duration: {duration:.2f}s"
+                f"Epoch {epoch} [{i}/{epoch_len}] Loss: {this_loss:.6f} Duration: {duration:.2f}s Mem: {mem_usage:.2f}GB"
             )
             if dist_manager.rank == 0:
                 writer.add_scalar(
@@ -282,6 +298,7 @@ def val_epoch(
     epoch: int,
     cfg: DictConfig,
     dist_manager: DistributedManager,
+    norm_factors: dict[str, torch.Tensor],
 ) -> float:
     """
     Run validation for one epoch.
@@ -296,7 +313,7 @@ def val_epoch(
         epoch (int): Current epoch number.
         cfg (DictConfig): Hydra configuration object.
         dist_manager (DistributedManager): Distributed manager instance.
-
+        norm_factors (dict[str, torch.Tensor]): Normalization factors for the data.
     Returns:
         float: The average validation loss for the epoch.
     """
@@ -312,8 +329,6 @@ def val_epoch(
     start_time = time.time()
     with torch.no_grad():  # Disable gradient computation
         for i, batch_idx in enumerate(epoch_indices):
-            if i > 10:
-                break
             # Get data from batch
             batch = dataloader[batch_idx]
 
@@ -322,7 +337,13 @@ def val_epoch(
                 dataloader.preload(epoch_indices[i + 1])
 
             loss, metrics = forward_pass(
-                batch, model, precision, output_pad_size, dist_manager, cfg
+                batch,
+                model,
+                precision,
+                output_pad_size,
+                dist_manager,
+                cfg,
+                norm_factors,
             )
 
             if i == 0:
@@ -397,7 +418,7 @@ def main(cfg: DictConfig):
     if cfg.training.precision == "float8":
         # we have to manipulate the output shape
         # to enable fp8 computations with transformer_engine.
-        # need the output to be divisible by 16.
+        # need the input and output to be divisible by 16.
         # if (cfg.model.embedding_dim + cfg.model.functional_dim) % 16 != 0:
 
         if cfg.model.out_dim % 16 != 0:
@@ -409,6 +430,14 @@ def main(cfg: DictConfig):
             )
         else:
             output_pad_size = None
+        if (cfg.model.functional_dim + cfg.model.embedding_dim) % 16 != 0:
+            input_pad_size = 16 - (
+                (cfg.model.functional_dim + cfg.model.embedding_dim) % 16
+            )
+            cfg.model.functional_dim += input_pad_size
+            logger.info(
+                f"Padding input dimension to {cfg.model.functional_dim} and {cfg.model.embedding_dim} for fp8 autocast"
+            )
     else:
         input_pad_size = None
         output_pad_size = None
@@ -418,44 +447,11 @@ def main(cfg: DictConfig):
 
     model.to(dist_manager.device)
 
-    # Configure domain parallelism, if enabled:
-    domain_size = int(cfg.training.domain_parallelism)
-
-    if domain_size > 1:
-        # You can use -1 to one axis to indicate that you want to use all the GPUs in that dimension.
-        mesh = dist_manager.initialize_mesh(
-            mesh_shape=(-1, domain_size), mesh_dim_names=("ddp", "domain")
-        )
-        # This is a subset of all the GPUs, and will vary depending on the process.
-        # Think of this as slicing the global mesh along the domain axis.
-        # It will contain only the GPUs that this process is sharing data with.
-        domain_mesh = mesh["domain"]
-        ddp_mesh = mesh["ddp"]
-        placements = (Shard(1),)
-    else:
-        domain_mesh = None
-        ddp_mesh = None
-        placements = None
-
-    if domain_size > 1:
-        # Instead of DDP, for sharding we use FSDP.  It's possible to use FSDP in the DDP
-        # mode, but since it's not pure data parallel we have to me more careful.
-
-        # First, distribute the model so that each GPU has the copy with DTensor weights:
-        model = distribute_module(model, domain_mesh)
-
-        model = FSDP(
-            model,
-            device_mesh=mesh["ddp"],
-            sharding_strategy=ShardingStrategy.NO_SHARD,
-        )
-
-    elif dist_manager.world_size > 1:
-        model = torch.nn.parallel.DistributedDataParallel(
-            model,
-            device_ids=[dist_manager.local_rank],
-            output_device=dist_manager.device,
-        )
+    model = torch.nn.parallel.DistributedDataParallel(
+        model,
+        device_ids=[dist_manager.local_rank],
+        output_device=dist_manager.device,
+    )
 
     num_params = sum(p.numel() for p in model.parameters())
     logger.info(f"Number of parameters: {num_params}")
@@ -464,8 +460,6 @@ def main(cfg: DictConfig):
 
     train_dataset = DomainParallelZarrDataset(
         data_path=cfg.data.train.data_path,
-        device_mesh=domain_mesh,
-        placements=placements,
         max_workers=cfg.data.max_workers,
         pin_memory=cfg.data.pin_memory,
         keys_to_read=cfg.data.data_keys,
@@ -476,20 +470,14 @@ def main(cfg: DictConfig):
 
     val_dataset = DomainParallelZarrDataset(
         data_path=cfg.data.val.data_path,  # Assuming validation data path is configured
-        device_mesh=domain_mesh,
-        placements=placements,
         max_workers=cfg.data.max_workers,
         pin_memory=cfg.data.pin_memory,
         keys_to_read=cfg.data.data_keys,
         large_keys=cfg.data.large_keys,
     )
 
-    if ddp_mesh is not None:
-        num_replicas = ddp_mesh.size()
-        data_rank = ddp_mesh.get_local_rank()
-    else:
-        num_replicas = dist_manager.world_size
-        data_rank = dist_manager.rank
+    num_replicas = dist_manager.world_size
+    data_rank = dist_manager.rank
 
     # Set up distributed samplers
     train_sampler = torch.utils.data.distributed.DistributedSampler(
@@ -508,9 +496,16 @@ def main(cfg: DictConfig):
         drop_last=True,
     )
 
-    print(
-        f"num_replicas: {num_replicas}, data_rank: {data_rank}, val targets: {list(val_sampler)}"
-    )
+    # Load the normalization file:
+    if cfg.data.mode == "surface":
+        norm_file = "surface_fields_normalization.npz"
+    elif cfg.data.mode == "volume":
+        norm_file = "volume_fields_normalization.npz"
+    norm_data = np.load(norm_file)
+    norm_factors = {
+        "mean": torch.from_numpy(norm_data["mean"]).to(dist_manager.device),
+        "std": torch.from_numpy(norm_data["std"]).to(dist_manager.device),
+    }
 
     # Set up optimizer and scheduler
     optimizer = hydra.utils.instantiate(cfg.optimizer, params=model.parameters())
@@ -544,6 +539,7 @@ def main(cfg: DictConfig):
         "scheduler": scheduler,
         "models": model,
     }
+
     loaded_epoch = load_checkpoint(device=dist_manager.device, **ckpt_args)
 
     if cfg.training.compile:
@@ -570,6 +566,7 @@ def main(cfg: DictConfig):
             epoch,
             cfg,
             dist_manager,
+            norm_factors,
             scaler,
         )
         end_time = time.time()
@@ -587,6 +584,7 @@ def main(cfg: DictConfig):
             epoch,
             cfg,
             dist_manager,
+            norm_factors,
         )
         end_time = time.time()
         val_duration = end_time - start_time
@@ -613,9 +611,12 @@ def launch(cfg: DictConfig):
     Args:
         cfg: Hydra configuration object
     """
+
+    # If you want to use `line_profiler` or PyTorch's profiler, enable them here.
+
     profiler = Profiler()
     # profiler.enable("torch")
-    profiler.enable("line_profiler")
+    # profiler.enable("line_profiler")
     profiler.initialize()
     main(cfg)
     profiler.finalize()
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index 129314411e..45b6410fea 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -385,7 +385,7 @@ def __init__(
             n_layers=0,
             res=False,
             act=act,
-            use_te=False,
+            use_te=use_te,
         )
 
         self.Time_Input = Time_Input

From 4f4906a8692d44dda3b63a8a154ca8cc56be0abf Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Thu, 31 Jul 2025 13:33:17 -0700
Subject: [PATCH 21/23] Ensure embedding is not None before concat

---
 physicsnemo/models/transolver/transolver.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index 6fcd0f5998..cc60ccba23 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -507,6 +507,9 @@ def forward(
                     embedding = embedding.reshape(
                         embedding.shape[0], *self.structured_shape, -1
                     )
+            else:
+                if embedding is None:
+                    raise ValueError("Embedding is required for unstructured data")
 
             if self.unified_pos:
                 # Extend the embedding to the batch size:
@@ -514,13 +517,10 @@ def forward(
                     self.embedding.repeat(embedding.shape[0], 1, 1)
                     # .reshape(x.shape[0], -1, self.embedding_dim)
                 )
+
             # Combine the embedding and functional input:
-            fx = torch.cat((embedding, fx), -1)
-            # Dead code?  Not sure when this is used.
-            # if fx is not None:
-            # else:
-            #     fx = self.preprocess(x)
-            #     fx = fx + self.placeholder[None, None, :]
+            if embedding is not None:
+                fx = torch.cat((embedding, fx), -1)
 
             # Apply preprocessing
             fx = self.preprocess(fx)

From ea6c2c91ecf34552a3bbeb2e2cbee84dd386a83b Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Thu, 31 Jul 2025 13:41:45 -0700
Subject: [PATCH 22/23] Update transolver for minor details.

---
 physicsnemo/models/transolver/transolver.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index cc60ccba23..a011bc006e 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -286,7 +286,7 @@ class Transolver(Module):
         that the number of heads must evenly divide the `n_hidden` parameter to yield an
         integer head dimension.
     act : str
-        The activation function, default is Gelu.
+        The activation function, default is gelu.
     mlp_ratio : int
         The ratio of hidden dimension in the MLP, default is 4.  Used in the MLPs in the
         PhysicsAttention Layers.
@@ -481,7 +481,7 @@ def forward(
         self,
         fx: torch.Tensor | None,
         embedding: torch.Tensor | None = None,
-        T: torch.Tensor = None,
+        T: torch.Tensor | None = None,
     ) -> torch.Tensor:
         """
         Forward pass of the transolver model.

From 9fcd89ced6cd35e70af199f9b5b27aa5fe1556f0 Mon Sep 17 00:00:00 2001
From: Corey Adams <coreya@nvidia.com>
Date: Fri, 1 Aug 2025 00:46:41 +0000
Subject: [PATCH 23/23] Update transolver tests, clean up lingering details in
 the model.

---
 .../models/transolver/Physics_Attention.py    |  26 +--
 physicsnemo/models/transolver/transolver.py   |  11 +-
 test/models/data/transolver2d_output.pth      | Bin 0 -> 117033 bytes
 .../data/transolver_irregular_output.pth      | Bin 0 -> 199129 bytes
 .../data/transolver_irregular_te_output.pth   | Bin 0 -> 199147 bytes
 test/models/data/transolver_output.pth        | Bin 116814 -> 0 bytes
 test/models/test_transolver.py                | 220 ++++++++++++------
 7 files changed, 162 insertions(+), 95 deletions(-)
 create mode 100644 test/models/data/transolver2d_output.pth
 create mode 100644 test/models/data/transolver_irregular_output.pth
 create mode 100644 test/models/data/transolver_irregular_te_output.pth
 delete mode 100644 test/models/data/transolver_output.pth

diff --git a/physicsnemo/models/transolver/Physics_Attention.py b/physicsnemo/models/transolver/Physics_Attention.py
index f6571ff37f..5f4b5746eb 100644
--- a/physicsnemo/models/transolver/Physics_Attention.py
+++ b/physicsnemo/models/transolver/Physics_Attention.py
@@ -74,9 +74,7 @@ def __init__(self, dim, heads, dim_head, dropout, slice_num, use_te):
         for l_i in [self.in_project_slice]:
             torch.nn.init.orthogonal_(l_i.weight)  # use a principled initialization
         if not use_te:
-            self.to_q = nn.Linear(dim_head, dim_head, bias=False)
-            self.to_k = nn.Linear(dim_head, dim_head, bias=False)
-            self.to_v = nn.Linear(dim_head, dim_head, bias=False)
+            self.qkv_project = nn.Linear(dim_head, 3 * dim_head, bias=False)
         else:
             # These are used in the transformer engine pass function:
             self.qkv_project = te.Linear(dim_head, 3 * dim_head, bias=False)
@@ -187,12 +185,10 @@ def compute_slice_attention_sdpa(self, slice_tokens: torch.Tensor) -> torch.Tens
         Torch SDPA implementation of slice attention
         """
 
-        # qkv = self.qkv_project(slice_tokens)
-        # qkv = rearrange(qkv, " b h s (t d) -> t b h s d", t=3, d=self.dim_head)
-        q_slice_token = self.to_q(slice_tokens)
-        k_slice_token = self.to_k(slice_tokens)
-        v_slice_token = self.to_v(slice_tokens)
-        # q_slice_token, k_slice_token, v_slice_token = qkv.unbind(0)
+        qkv = self.qkv_project(slice_tokens)
+        qkv = rearrange(qkv, " b h s (t d) -> t b h s d", t=3, d=self.dim_head)
+
+        q_slice_token, k_slice_token, v_slice_token = qkv.unbind(0)
 
         out_slice_token3 = torch.nn.functional.scaled_dot_product_attention(
             q_slice_token, k_slice_token, v_slice_token, is_causal=True
@@ -255,7 +251,6 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
         if self.use_te:
             out_slice_token = self.compute_slice_attention_te(slice_tokens)
         else:
-            # out_slice_token = self.compute_slice_attention(slice_tokens)
             out_slice_token = self.compute_slice_attention_sdpa(slice_tokens)
 
         # Shape unchanged
@@ -263,8 +258,6 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
         # Deslice:
         outputs = self.project_attention_outputs(out_slice_token, slice_weights)
 
-        # print(f"outputs mean and shape: {outputs.mean()} and {outputs.shape}")
-
         # Outputs now has the same shape as the original input x
 
         return outputs
@@ -280,8 +273,12 @@ def __init__(
     ):
         super().__init__(dim, heads, dim_head, dropout, slice_num, use_te)
         inner_dim = dim_head * heads
-        self.in_project_x = nn.Linear(dim, inner_dim)
-        self.in_project_fx = nn.Linear(dim, inner_dim)
+        if use_te:
+            self.in_project_x = te.Linear(dim, inner_dim)
+            self.in_project_fx = te.Linear(dim, inner_dim)
+        else:
+            self.in_project_x = nn.Linear(dim, inner_dim)
+            self.in_project_fx = nn.Linear(dim, inner_dim)
 
     def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
         """
@@ -323,7 +320,6 @@ def __init__(
         self.in_project_fx = nn.Conv2d(dim, inner_dim, kernel, 1, kernel // 2)
 
     def project_input_onto_slices(self, x) -> tuple[torch.Tensor, torch.Tensor]:
-
         # Rearrange the input tokens back to an image shape:
         x = rearrange(x, "b (h w) c -> b c h w", h=self.H, w=self.W)
 
diff --git a/physicsnemo/models/transolver/transolver.py b/physicsnemo/models/transolver/transolver.py
index a011bc006e..88a5bc1ad1 100644
--- a/physicsnemo/models/transolver/transolver.py
+++ b/physicsnemo/models/transolver/transolver.py
@@ -220,7 +220,7 @@ class MetaData(ModelMetaData):
     # Optimization
     jit: bool = False
     cuda_graphs: bool = False
-    amp: bool = False
+    amp: bool = True
     # Inference
     onnx_cpu: bool = False  # No FFT op on CPU
     onnx_gpu: bool = True
@@ -423,9 +423,13 @@ def initialize_weights(self):
         self.apply(self._init_weights)
 
     def _init_weights(self, m):
-        if isinstance(m, nn.Linear):
+        linear_layers = (nn.Linear,)
+        if self.use_te:
+            linear_layers = linear_layers + (te.Linear,)
+
+        if isinstance(m, linear_layers):
             nn.init.trunc_normal_(m.weight, std=0.02)
-            if isinstance(m, nn.Linear) and m.bias is not None:
+            if isinstance(m, linear_layers) and m.bias is not None:
                 nn.init.constant_(m.bias, 0)
         norm_layers = (nn.LayerNorm, nn.BatchNorm1d)
         if self.use_te:
@@ -496,7 +500,6 @@ def forward(
 
         """
         with torch.autograd.profiler.record_function("preprocess"):
-
             # Reshape automatically, if necessary:
             if self.structured_shape is not None:
                 unflatten_output = False
diff --git a/test/models/data/transolver2d_output.pth b/test/models/data/transolver2d_output.pth
new file mode 100644
index 0000000000000000000000000000000000000000..031955ce2ad5e1b8d2ac3866e1ccec91a71acbbe
GIT binary patch
literal 117033
zcmb5VbySpJ*ft6%A)yF}(kUP!Eg*7TA_&qT0@6r_NQ(+cw~C0^i728Xc3~?PVt2P<
zch`A--}mQv-?h$}wPww%XJ+>7*|Yb3-&fpzp1pdDh)79^{J(DMA`&7wS<z|P>8ZK#
zS(b58=~HqtrsSB#Mdw7DW=u+TbrKT^4J{U%ktyOS5}T8r6`N!lH6<rGHCs9=D?VmQ
za%x;uPQ38(tf*W|=}b{0qhLwlOR|M`MkmH+iv8b@NV}${N9T<HzZa!5dj*RM?=jEp
z9o$QJjLn#mDenCwBPUa0oTrFqZ|@Q6o}yKmlK<!Nzh_c4v9al?sqwKn$?0j?(!N=7
z@mcY4W0GTYq%->r@z0db$&~TWl+8Bv^OTetI(d!XI$_lR?tcI4&(;%>`X7I`X+nC|
z<mj9zVJ)(SFLQP3E&Te>P>cUY-u>=Q*?;r?@BY7!tRZ6kKaLzTJtsc>znQx_ng5?L
z{~P!J>n$}oCnq)jze!Z!v5JH+f&cFR`y^CFEdR$z{P(Y#nY%if{{Ngo!kx!3`dH1v
z?TS?2Squ4t-h4d0j4iWcSuj4B6IQBlP4Rjd=M1Edtr(kZB8jx|ymj9h_BD@CBq_?m
zv=vb3ZA-I#Ct!5GnlY<f_-Ar1Q;f&+wofo0_2|Rn?;NN^rt|lsBHC>x>VMXvPe~{n
zeX4lewv?yW_+hP79%miU#P_Zm?CF01>W8Or`8wgX9yKcWy@w*DeoRm6gvUsIv^D5a
z+cplzH`L(q^D;<<X+w6T3L0Kb=2Y)+UifZ}4<$wnN^#}a0gYJG7|Pdk-6?;hhzFzn
z*!V*W!%m0cW#|-mebeOXCTqxF%7t>V8%mW`S!&yYqfeLOK#dCrjl75c5+?X)>I47m
zgE*?%8*VrEz~pK$b>tr*Q6(3yiwD9bJ)9K>1NbVng1Z-g#f^XzcKOuMc2h02-#fFc
zqL{G>9VoHLAj9hL*;SbtHy+^3xZzZtS%822eR1@BG0!cJhw`tTn6M>^&(}?-=ihSN
z@0ZPo#l<Xoa1j+XTC|$yKtsd9j990`{hkwPoU2EB4Qp6ATCqWUAh)>(P`l6=2U9(e
z{6&-nAEMZ`?Jv}ul(@3OmOEp5b6ZL=du&Eg$#Fd9>YLI_JOH)h3~{OZ4PM=dVv^H*
z<R!`R<Q50cvR2|fdre-+m4W<CVP3~V`Qg)I+&&x2j}hI_IW-skyOJn=_bQy91v4VV
zg#IbEl-l+I$KMZt&OkSY?5o9_L4*0w_YJluno`#y8u`Upj7!Mk@aDdZ872;o&-bv}
z&xnn4Y~kip4n}xk$;%(O__vxqwuI!Y*(@%;fx=(8eDPos21$<RHS38SmDPYp(<Woz
z>J}8#@50-t>CDk6XXxTGT35ZmgO{R|Y_#CBZ&fr(D(5a86*k*7W3{{#gCeK%fnGUH
z=7mzEat8fPCi6*^FM}N-=$F}x{+=ONdht5a-cREDs=-*XWH8nCkXv3_()`O@@KrMH
zhKIsMR~y&T<oPhYm~{b3IQr2Z6D;#-krl>Y69&Rib{gJlmN2dPEdJQF!r7^WQrG$-
zW1R<|?O1^0hKta#v50f$zlG9lWrppVi@MPssFvD<0Q*Z=r5s9SWTK_|6ar7=Aa|c1
z7Nu#k*KTq2Z}-Bn4I*6h*^*^Gl04DR2RSm6S-#koHBxezY#K(NstT0(#AAMDB90fX
z!c*_rG%weo`^p;r2s;Grur07!6~S<gYFyS?0LQ8sT&hrv(8x+|dF98OYbx2+Uci=`
zAYM6LMM)87{Ovs(T{~*1V_eCg#ih9Pxr}li`dl=#1;G=}!Qf9Jw^SQ&R^&uF?Awl_
zH<Q@8<PWY0W4-!P4(TD^&@KKF|1SG-SxqOpx{DDOo5^pR##6=k2oA`P<=2OR_t|8I
z{?=jNEob34Za%J<P3F7zgV=|0oHutW2A>&>fifex)7z07PcDM|6*<`ED$?PGJ&qlA
zr)`Q8mwZw{gPI~2m}y}B#JBjK@e#(_dKAmshN$X6Se}wV1G9P1Jv9|+eFq`ujvRH~
zt3f9(2kYOMp!b5Y_&Ui1PiEeL`qX6j_Zoy2lOAltF09yBg)`Z!@bSV-el&@M+sN6R
z=wHSi8wT=5vlkORmH43PJ}yno#9>)0+{_=#yLl;$Q@n$ry-K0!a0LFdSHWk^BlL@Q
z;pnb<D8(f3!I5nIOxTWH{*&mpa|XXg{X&fFJk&*4Fll8u{U1F<#XKE)JeY^Sr8VH$
zVrHvU@I$=`#Z3?6#Bz0f8ZnKRUVOskf$I2s(-;k6ON1C}1ixx63|hPpnS~LskC}xP
zwTrN6s59C-6R_@)7hn2~=f+LS2r(YV4L6qJrM5rLZ61!Fnc6tt8VuFC5$v}|n<siL
zM&;rZN{^h(bkzZ{5g&*QcNGpFoXUeM^y!wNK;`8M=#>AAN5vM1pQVIv(n|cKY|Fd0
zmgsXU7hkR$qCvq9O$lOXtjR^&-w3*_^kL-u-aKDY%_)U`JTQ9(6*tzfj?H-Ys{@*v
z(X9Tt4kt(Eu>YBUEK=--RzNYYe>{u_uA&^(Qp7{mwLGdknmNWH==<&<YOC&`;m!cg
zK3&P{cGW2Meu{~!Ct$!XWwz;8^U<DZ+_!Bsl=>F4d5bX$-IXcXp21%c=^VAOoX@n1
zxLvC+XFO6u<AqYT%Iw6P^oi`NYLCd7{ULY!I?izkVk{I<x<n0KzKSfZ?az<{$!r{Z
z8JR=NxiWMCik~G5|7%~ApH7EoFIj{u)ncUcWUAVlvg<)CUxxSQ$190&oEk?{i6F|(
zbK=<0eR!3Zh<W!^x##p#NT0FCSKHxeJY0<MUc>pjUnnErj-;J&e~jC&iox-{pzJLH
zA7?p~C-*{qkp*&o^<``7XwEE8=l)>@d{ABrt;1>b*qKS=3sWffaUVAH1_y=qp~<pb
z9@{>RZbOzKV(@O{NxAaDf&JKWp^9E1cQDa_F#T7EwdD#74vS@N&nYC#o&mq<l{}HU
z2x8)q%rDe|mtG#WOU2->jUR8c-G|*SH>?O<kI5kokiS2kiE`O2%s&OJo=m-i>71cm
z%<`xT#IMQYUbC5~70&E6hi<s4#<9Ltnm%7OnKi=-J7hfRUhRbAxrGe-8_P?3zF@7w
z9-P>%g-?Yeu`1P_E>{w8azQgjUsb@24bdzUe}zqpqnKGN3(0OH?p;*CXq^s>Ild1C
zjSjFmug>RL326AI&T#%ln@Kc_o-Bg2x+Z$QOTo=)H|n47gHQ51boEW)ubO#qUb+gG
zUe{ronl68=xCrUX0dQQ`j{Pw;q^CFaRDwBcX#~}lRG>?&ls<J~JQH&OY9lK6YDxeX
zmsMg^SUzt)uVD6t6wGt*<)T*}Y%nk8y4#aD&HOGVdQ7KDQ#tDO5-B@kln|rp;eSV(
z?MvoiTpu4gtV`gk<K;}hw;1h<D)_Z?1`FS+!BqPmE}EsXa%LXyCq}@dr-YmTT|`N^
zEen0-AX<n`JzCl5S=1jf;w6~7QXdl*<nlsJG5d-Q;{LB^p!z6~?`1AvWojI|J7=Lz
zZxAGFHE^J97QS!MKy<YVQh)ohVTT{2QU+80WE$O~6<}wjfP|^ZlsYEHIE@IdbuNXn
z5KoT<`XMpD38MX`aG`_`?P>=hYu{m9m^qpq@z0Px%!YCIMzYq|g1>d=p#EPOB`)>n
zmMRIpZ%pJZgCLw|Dch!3ah9ep^zRoyWb<UmKc0`r{af&ATN&i1567hg+H9Vdi=(p?
zd8W>U!`mwu5pPaiV||vb9Ksbf;XHrDpOq`8^Pxf@4}8sG?c+-@y?Y%kx6=7{ZXVD2
zIrDG*2xM#M!TnA-KTj>dywEe4@n#;XW5+RNXf>Q4IdhOo6y~)!@UDCgPtB8OSyKvX
zhpVAXRmgSOQ)xP&FVfTtFd||x4(|5H>Su|(JI@KfU%Jt6YcO`Ht8wJ*H|V(F%LTPc
zw2%oxSU)3{jJ%JBZ-f{cV1U?63!WYl!4)A*xcmMb9{n=ru{L|!Ok059STUaabr6q!
zOXA{A6P^|~p!&n_*rlh7p97PT|Jx8+@A}en>sp++H-OGL8xg%WhntrfQlzCn)j!4J
z!CWt{+SHBv3$4+)wT7t+i}+_!It|L+!0W&V^ieyBjLe~I$!x@hVU;XY)!;QPMb@-T
z<MKDhFmaI`R6=)vQ{4G@cm+3H{fN2hc`V)}!6z+a_*teCt6m+)^MnX?sX4N?!V;}@
zg<Nq!o;x<}L#@q9cr+#Ob@6D{H4KN*0Tpz)nb73A3?CT&fXr8KCJyL<$X9o`N;ToZ
zd1n-~y0RlGm{)eEV(NHFj_<b@CFj54be$5;br|xsTP$BGhO;~PHJ0g_AlO+CwZ21`
z(dmK5Z{Hzf$vU)`4`N5cb7UrEplIDv<gSjwm>trbYUs*pF$?52<>RxP1t!fnh6Za(
z`aL({&hO$#3<!jMxj2K47-0VT3O<OtgRDJZ<Naz@YDM#Sa}5>l&4#S58f_NZaB@x)
zZmCC8aab+eq#_`CUW9$)zoNXSg6~p1s5`Kn2a31C+^&?as@tFum&hM6X;>K7ghK`6
zxK}zK>N{O=;krGeKCH+4PYGOd<26PkUO@7X@f<!+hWCUu>aG~YUG19uP$=xzj=31K
zr<|qBQh5GzZyvU?V9$kitcvu7M|KE|yf47>Qvt)qhVZY32{kwrdt9Ewv{Hi4ciN#n
zB8esq3D`QfKYJaJrgLEgtJC|kaO^O)FN>hOBT#To1;JWAP@J?B{+IkYKFFPpDz?0K
z;{{e6Xhzu%Nvtp2fZOXt*kNjo^6V6r_#TJ8dI<I$7NOI;BWT>6gMDJ|NZm3M%U@g4
zrG7HhH2ipEZaNPxz7CbIVYu_a8vz3YIkIv(zqFOn)Hsp7gt%a6kjNoX@;p4biWg3~
z@b<8|sJ)wvqeb4-{yCD9n#W?dO%%m%<gjAg9OU*Ea!ANEIMg|Dnza@VT}z^Z+60d3
zDdx}2B{2W0PxGY#OmQq?n!|GxA{oP1lGaB?u+Fg`JEm!}AuNbDEG@Y1=yYD2ei+G5
zJ|aQu8scXEh4tzReE4Al58HM4B>NRnDaJI`2|;*v8h<a!=2eF|7&Uqnd)Xd@!go!&
z1vjDZ-*Nmo=rUU470~{>2gS)s5F5M}i+h!$*fWW`mxHl%>Njw`IL@V8F~4IPqFv3{
zbxjGg?|Lz!z85C!Zh-wq2k3{7rh`H~{=T>4(#b{WS-2hj)(Crhn<vZlC!pk5AQ!!~
z#T{Y41t~0mw`Dk+E><u`XDr`jR8sH$MEVrRbCp9WZ+SK&V%$AsZCj3;Yac>6u8c=b
z)wp2bX{@z~=aAj@2sRY(m+f#q@O8%UG4}MYHAT`z0ry|5fun+eW9Aq0+hJ>p?J480
zBSUz!qzQHr)39Eul(lP}(Yo3fx+8{Sl9nL~a(Zy<rz*ZLoJ!dzbCH+o$@h<MV~(Z?
zM7+)5py$tzDkt#z>vAM4v7xx_DO8%8!Mebf4cGlCz08%hva(349E+;h5Nf2Y!JLC7
z_<sKr6y3bB+$)GjWHnH^I|b3>wOFE;#s&{<z83I>ghR4$rV_Ex#FvuB)z~Mlz@RO9
ztiHPo-}D3N;_Airqy3P(%!f*E1bpO6c$gnT*8DIgJ<s5cEv4|vbb*}PVzey#h9wR~
zG*h>s*Pa6qS)Pb)UniLRi1VOm1lz}YqJ5$y4{E2Qk9-Z!?ysPE>J-@TOyi2@PqFpk
zM0(vE2(_b)DBUQ^6Ox{+wVpsDc_WtA)nLpD501WI!NEd38eN^tm2sYkRq4R9E9G=^
z8N-aK&3I{Lg4b>#2rMz7>ftyPR%f$ycM)Z!RPcLV5o;Gs$8V)cu-W$$iyn-{>P=(e
zckDSF4XvqAObScRHBQIj?<2w!ezn+WD8;@TIx#6LoY}Jsuu&w4m7)ug(KnH~E7zdK
zO&y(kyYOe6Kc&P6VCcZ1XpWXb#P@L=<30eYOU~ld(?sScwBX3aF$_6>TBrdJLnTRv
zQXy&(&s%}!mbtj0br|<&WKj3SO}r`Fj*u6T)bDo?VYkF+*zgM?Co8z5LW@TamvH3v
zG<w_5hE2s>_%=@Fx^O9;dN7KO31g|2v<_z)0%$kHgxU?Iw2_ja#Na&$T_{f1kFktw
zUWm4XmQ*=C3tg3Va1+jHhfXmp-Vl>oCvi_de}vw5rRJ&uY#!Z%n6zScmX2hYj1PSs
z6llLaldl%VVnI(dgAW~ty@`->aTgtDRblhN5$d}v@gepO?!TW1G>pKPb5|iRTFkXe
zGq^ZQiy@X$wD~-Y>8&o*Kd6Pey2tn=w+C`9QxS4j9_M6TA@MGoQcuUT*3<ycKAST5
z;8{G&a^Wm#Srndfg5uzAoE;;}{GD>BE%M=({zm*geKTGs55t!34%E*agmVYy;KK_s
z-cDW%Kl^y<T4k_aI*3mme#EnjV<8o~4Y_g~Fv%&1X9M%8s5p@dI;re+)e#*tOF-F5
znig$^s$W08v|WXy`_7EWOh&z^Fpj<w#iNEnL~k}iy3=@~rx;sy+wsrpO&HLcM~R-3
z=u)qQ^Xv)ykWY9;wIH;=0XkA8cyhHG4Kyxd&)$)Ea@UihO-bBwLX4ZUmg0e46guz6
zP)u$Gtc<jIrNMy8N8IT1NDp&v59OP2aYXHCUY#Gxg&hXC(=&(*4!Ppql5m)*rE$~3
za8A?gi#<d1@H^U?TTCn|w><&d9Xn9<eJno9_~71QG4|dma5TN-*jK6>Q}>SKw$lq?
za4HBN-0LuAYZR}^>_*C<B$yqk$DOtNVK5<yQX;RR*}WTz8JjUbYYo<$I?(D-G;hq$
z;`jIa(I;R!C;sz=^_*ac&zZq2gBs4MYJ%(&fg6op0hNi;H1S`C4Ob7tGxQ7M<`;2^
z#Z}y_pMgs|X7KvUQ_!Ec3ga8gm?K<2>l@F8<pxN5>5DUC|6<$kr8u?1jt6_nxOjpd
z@95par)V&2bPPM+EyBkQ9ng2QfM4HS9vONL^X_Z&m9{2K-T^P?>cRS{4JAF6!m586
zwNn7)BWgS<K9C91BDj8F98b1R!HYX@P_J!@*<UlcFZCi?jGkgc+iZjtO=Z(+KWfhW
zh-R5WY+jtsD66A*FvJ&E7M;V7?E#cdu*8e#Q0Cr^qvf5I*#5wmmbau}wa%F_b_;RK
zN}c|}@6fN896p?o#j-bdQ2!~M7f%Xxzor`eY<<}AHV4Q1l+mnvAO4n%$K~EXQ6?cn
zsa54XvZM`CeWx&2z!r7O-=p=F85*TsDAi{@I$!u>>C-{f6JkwXUpM%DU4`J;S^WDg
zovzb0sd1u&s{R^08dt<G){k(wxq^PNBaympD97~Oi%%BDywW2|<GNb33H*b4zX}Gg
zl;A}3G|mp`&EFT5@n&-;lGAdT|1^nHKkdbd7e#zoc^`{n^XbsH8=5T%OxV8~$;VdX
z?W6%%*F6B5x?7QbO_CxhBVqI25H2>}+%NG7QF_<0p#K_NQXfVKLs800OK`kUM24yh
z-1%yES{#zbHeX4O+20=v3$9~?rw(N{_2R_<bxxbL4U>}m&^~T3a@Ko5(Rd;1Qs!a$
z=Tm67WDcLoI&5uJ;;ugOxEg-}=FT(ld&E=(+4Z4DttRIjvqo~|T!bEfhg-S6JoCz)
zmHR{K`*s6D%d2?6eL8*X<GDKj5hO$Ub9JjZmn|3$%Yq7e%(CPm@p9(NYcb)GKh?}D
zxuDCB`ByX;SF;~+?pI;%D6Dme5eC{`!O1{b+AJxCs)r^f`g})2?=|qR8-l%os+iZL
zM-R92=y$XV^&5hj@m7QW0jDrzg#=&wEk@y_Os4cJW%xof+%a$yID*CaGEsy#_Pg`S
z=3pFe4B&`9wb-V-3bTjZ!LQxhuw^f4b}0#7+kZgwN*DU~XmI};N8ae}f=)#-;v@G%
z`=Tx!2gzcI=42e)@5Z*v2~=?y%)^@^nYK$GW<_p%7T6oQ*A!5CToGN<%Avbahu6kW
zVC;4^6t5Ndh(jZ|=k{$Ryh@|eu>Ht7Wq?z@LwUG6okJByQtIXzblvIA5t99wr6tay
zn@;qXJddYxdf3-m&EN}7xI9Cf2Wlqrb$2)XcbD_nmggv{P2naXzke36z@X+8$S%xe
z_gHWI{%XK6QKIZIJ%T*~-|*H=ngJIYkmVE42R8z7qjM^+O;tzszdYLa9*Uy<9q14l
z&&TCyeA79K;pc*QIK>v(Cxf_M;K72oUPkBWg>W69N+tJHj>>bRh4KKpG*89PL>a8g
z8pFsmQ+Nx{Exmiwrp|(U!#t6p>w<rA3*l51z{Ub699R|tzj5Q4wLlf|OT74HsVvM~
zjM%$=EIaoJ82pzWhd%6ril{HPsfTm%jsDm@YY_H{F2#t_ohYl*X86G-cxV`5?j=iD
z-YntgRm-t{PY6{vYS6@P214(KvD0Z7oT-PNzN2tA=oHQeNYnEEUQ{MF;hf1ITnMdZ
z>-;#*{=Ej<J+tAfZ@`t^FJX1`8)mG$2vM;Fz8F=)n+YXcu3*C2(GvU?aU9y;1s=vH
zoAwsp@htKMmORlFFp59hN`^DuZ3I-Dm3cqHlH=YC<Eo}fIAN5>YikL4v2a9rdcwrc
zPdLL02<m!<?(4<~O>M&gBPCR6_D5`U6g^!X*!1fR<OQv1`jTlhId2G9da!$I1@^rQ
zVyCSwdk=bpf5xU*xn?h@HHuHZe}~<TmpIYb8?OJvVYw<0-5-@0I{h_zH&|nl$YB18
z7Dt@73+6Vu@%3L<O7=Kn+Ru^D6?u=FIx}!s&Xa{TRctD_jc3C$pl@MK?ScK-Qdi9-
z1EkqCZU_9onc%#U3#V=ouyBqYY<foVv{nU8HUeiS3ALBsQCyf1$^ju0XzYFxYE?Ph
zBQDOK2w@ICoEUlj5bQ0Ac~Hs`8%mbpWOx!)O&Z{7Gla8k0=Yq9GT!bQNtrPdVKQzQ
z<^?arY{YW&sR7(l8iB*2C9EG}!<`mGS-1HY`ZE~Q<kn(ivpC<%6+qWa;26f5v+kfh
zg1RI4^!sR5sRYrZCl31$4@13e0?J3U3-7B&bW92@ZRN0Oup*>)6jEv2WgP#K#vP7A
z{Z-Nf&zz4i7rlj;jaK|QI)MQ$O1P_P$nX8dvHxxm+B*AS>!Na0tA?}wk2o!}40v#-
z8Q<M_iqC(aL%zxg?^hRaYMThJ)|OHIHw9k5l2+=0+<x7Vu87BgefOcY!HDKA(!A!~
zhcfB2aOi{{1LsNco={sg{ETJTzi6802>PazHLmA}LvK|cUuN2(x^*ZPxsAd)TW@wg
z`vtuX>D(Ps!c$_qAi2?$X<It*!pD*u7gfOenh~y@*TwSdC3t)7FFw>1v+!3wUO0V5
z@NYvVc+{f&h$a2cU&o^eb(Xh%fuG(cEO3#*cbAdKzq1SlR*}qFWdOfR`>^YSA1Aj7
zoOa)Ep86}t{L+(1yXDNnrvvF&tINwGefh;e4j(%YLTRC(-7QO_<%B@GH;m`6L<#Kw
zDGP_h$B1$=LRZuV6zYAz>XZ?fzgUlUVeb(h(t-fraJ;?H8&_s!aOZDbdL4J9$ngg3
z_x8kzDJ3*LIf_d>(oyR<8F|48oYzN-Eh*n2vScZI$|7*Kj}qHA%W;8m41Xp_a@}if
zj{d35$LGp8U!;nAmrdc8@jk2<ZN`Gtf6(z-1rjcLocGO);(7Ue6J11&fdzbcW)NJ@
z`_MQo95bfLv#q9_Ur!g~L7W=KKbgi?J9RKmz^Xr|M&V)ER%}_l1@8-uu>Pnu17?lG
zy$j@lMRrtQsKK6MbG(03hihX_;?E3Mu5=%TeNB!`Ym<gpfIP1X`bltMKe{9m+6v>4
z@N_g<gPpj0@(eB<8p#e?;)X#fWU?Z8=1?Cr>FcuhA0^Iddx7g66By_}gVA=m9GtTV
zmO8!hVx}%<)bz&MG-DP|aODYecYX=0#-iPFn0at59P5oy6<~zS?;1SuXBw+yN7LiB
zz}<AdK*8K(E@`o)M8tHul$P_?-Dy;|c#8!Q)vOaZUj1n!QInfY9TgoMb*kj%*Xg|Q
z(3+-b#oJkqRJgkn^(v)wJb4qNYo=0Lq>PL7l5p1ZIy{H@Fx)Io$n&bGnw7`4VX+t{
z`UPgYN~rCpOV_M&_>O6Yquxx|X3Nu9`U9RD&4EP`c{p5*ha=^oD=W&siXL?Q8btLj
zcR?o!WbiEm)H{ts;7lVP?hI!1S`B<V@4>H;1DL0hLy_fyoR_P`&lxrdI=L5SMsbLj
znFNy}6Mk`6j^~QOl(}cikAZ_RCM1|j&Cyi89LXL3H2H4vNqD6Aqi}FEFA1^KYl8>1
z&ZhI|!`t}zcmVhR*5v4D4QA`xvvhwSCVhF0@1v)3it|Wj)lBB^nbn-UOpedqmh=14
zLIy_dL)^7l$gK{=k!2$(+nj{*{pE}=&%qC48@%3p6MK5)Q)1u%7|oR7w#&&(N*%(h
z&d;Fz<{aGhb2#v2EU$?9q9f)4)Yc^O-{TzKlS=2Qr@zqoVK??G7vPY{G|c*wN}I{f
zEN3#bcTDA=hk6Km8bj%=3RpE&7g9$I;q*=m#<+lqx_%UYGLBcD+EFPYgnc{8@i$W$
zJ;I)L_>j($fIR*@td5sSeUP$iJ|>PR;;<j?c%t+N)6#~cJ+KdypC3TW88eD$NMnfO
zNc0ikiS_|f*mZ6IX0FQN+HhASbsBKmu1D}`_QZ-slkqC!DAwi&Gk0Vpy}q5szJPYD
z%n{<j;n7%>ug#6yMG-gFgR5KxZmXsRdskOd^mjO0jq^BXW)`~!-oUDmUzjc8!H%W_
zI5!{@e_G<$VLp!cDzjPfRp6OyO(B`K0ICNApnLTLCNG`Ny!{?9Se(h;h9Z<c>B6-x
zLVjwGhD1XFmGT7q<vs%MI~168<`_Kk40%qo5#dskQBj+~pe7Y8xe$rr!d@Q!T^%jm
zeiVrwgYvtZVbbpkZamfH+2ca3xAzl1EcuA?@s3<&p3VC?_83?Y!$Jvn1|Eq=^N&67
z5D}$rR6m?=@Z`+xZAflV=bcew==x{~C!UvLOpq>oj8<Xp##Gik=|ictvvI}sBP5|p
z)73frES$f<>T~$iatcOGi%}<Uj_mX`*yy<m5994&bm=%;vb@<CX+_VL<7j;$g5^^T
zF#bmx^+PQfSuV|=flk~bA*|oANRAmL%0nN@s90adSvtnp@m1ixOa}0Qfa4?j8zDU{
zkVhoS*m!C@Rt^ziQIi?Y?D3|3s{}v2T?1J~8+`ha$b)m-_}keL^MjoEw8@&s3={de
z&47=t#c|Q@<?wkDPn%`+*r2M;<Kv=nD0w=o=7;cW&kSBT<iV}h2VgG|#yf9U<EnrY
zcfA?RfZ`*9&X~vd1+f%QDB$yMDJp(d;;}4mL_OJ$l(q6W^vsT#cXvaw-wV8cErrhx
zZxDaB8I$|>=fS7-kQ{5tou^W0DEk|MN-F3#A(ri<HR1K@00LVja7fRa6PI2_#KK}+
z8#NpkzD{MV&vPuDufnv;wYXX&!9}U*Sk<M25&yJs{+Aynq-)avZk$n;fIXLraCw$1
zC*~J%^lTlLw8zu(mm%}k?M3>r3~t$4jNC{KHtw!Nr*c0ks_eiueW6CmHOEnNJ-nQ^
z7^zxWY>G-^-~uyFt_tR<Q{~j|P=`<ZRt#v4;g%((s1EnWiWi#PENJAGSc6mPwuq}-
zjx$2si|CQ3R9qY8^yt9Rb_~DAItqTLw@A3D2GK3bIC(o6+Xkt!P<$8Uekq~*Y$Lvv
zS#j#WTz>A~jO|IH44)~S9kVAek1@y4C@bc!ibSHqFq(v>B0|8J_HA#WH*yXtYitE=
zEk)1+d+=(SJ8lGLa$$Tk{$A2SVYD2-O`eXYD`e3%K^I+XB{=hr6jF7ELlc7q9It_y
z@dMD}QHrCJA3`(Ik?({1@#*dm%r}*X$H_d1&M`%*VFE6_Oyu?&d73M9<7xh6E?=KP
z{dPAv3-zk)Y(cN9If$0|X_zD#gK2dI+%bgMt}4PtjQ~~#XW&p%KA$v(aC_~1>|Wr?
z$;Za?(<dQ!s^;=prWzu~%W=ywak^hJz>!NUQTa2ED{q~`#ZB(`_G%<u_X=7;*fdsn
z8}q8nCb&KtOv=o`N1bWxH&@_fX4tY|!enZ82wso%h15&fhlFYJNUoj1OI|~<=B_5r
z7OQgXu_@ecWQ`45hGXa*U5F}a;NhGYoNG^}h^svN99KbcYbTDJk;Zy`4PH<^kHf#-
z!8oli{``z!N}f7@SM`CDl`|gJJ7AsKbS&`^JUKbR9N!qk*rQi5;)X0OJEZZ;T^U<n
zi(qn>2zFXW!snbAoZLik#AF>BuG-PPC7mWA)-ZXW!3cA3i*^;qFA~-uCyK{BL)o#|
zn1hPu!Xf4pBnB0+Bdr1wUq@4Enj#BjoVi*lhn>Cr8SfmzS>a*u{u<BC@4)kmJ_(w<
z9W;*X;8fUBJU$Xeha@$+_~&yj*J6S~Hf5*pz_#+GXnW;Gg>81QsL7<g?|LX66I6AB
zg=oI(&bO}|@xXaL>H>suC92W$VE`Q5J+Ml?4<|&2b6=^PfHCy2OfHMs2E!0HOdbYi
zm+<RiJs{+KrvXnO@z;um3MD*v=K&n%dE%S7I(1i{#Q6d}*6(|T8(s~F@IHgq4;gGv
zN<fC1;JH}o&auPla4%#j?rYod&A=DJ3G2iS*%DZJh48q%;OjX#6nj0TptV#C3C}lS
z={6t!RxaZsw<FLNpT;jmA++t@h?;*<{B`*!Hmu0uYQcYaX;B;<+f*pCrW6n2J8}F#
zB^7=H(oIw7;}Fi~RRifYtPQ3q>8wbM#e0D#Y?b>CgZD>p>6kVSo{8tPgu_r2xIe#h
z$`JilD&X<aP{@&_%D<V=E|^XyU7@x#@kd0N7^TL$!`%8F5=_Ne(`y!-jy^~Esd1Qa
zNvMkq7vq{*1hnd9Y5G|gSH~wadh;ZH%_zZ%P5La%x{4Y#9ZczMK*zJ-B7Y&rKbFRF
zI|sg*>c`4QQt)tpjxQgFV)JcF?6I<<a$SFX7C5cU;`bQSuEuS{?_u{vF+9Afho(bT
z%$oWHJ06K5d6Fo{j!0zbRz<3q4#(6%rO3P%$i2hmI6Hi(fC=@{so908ic>gIZ8kFf
z#JI7moNMkpN4!Ta?Pdh<?8Zb^4;qfo@4NA8(m>uXb>^E3=?oDx)gNzUnWt`qkM|s~
z|584;?M=sxzLAhv+lR}Vg7`~W(2a#UYxgE^@P`i;r{r>9+FzuY>%dYfj75$g(YMkc
zIsab5U+_|TpVwtWj4|sg3OMx65Fr=jvD)|#`t^9wK;i;YWt#=PY&_+!&clb#ZRnaf
z1)sau<I0Fk#+|fg+^s>plB~nMy6sTjv=s5#v$0FNKRY)M#HM_84qfa;lS5h@+II-8
z`aVWp`$O0)bj99VF1)l;jH`C~vs_IVZPCWqsCEWh4iDl>=M=n?6a2YL!_Yr1i9@$l
zVOe+|K4|u2QT0>Y(UQTMuTD5JU76e3*CHiqGs4<zDS7*az{TE#*2Q9;4bf!hiQBj+
zS5B<~wtRoy2Op-bg89V3$Sad!^`E7P%Sq+O-^*}jcscZj74uP65$k>mHT$kgxDJZJ
zYGXksOgDnKa;cDe@>m-a!h^f}3to_cc$pH7F**6vI3uiMzM$m^dX{x5FtTR~&Gu{a
zNc2DK{}YUbcZ{%Ofv`^50zWk0j#qc*K_f5~zdpDS;;OJ7zW_H&MsuvvW=LF%#O+I~
zQI+FCna54gWg722n~M3rnqe_lhvVF2fJjR$)c2>sHYa{hQ)0NoUZ`A-qtw1=_Rexg
zV6GEyZ<XOY^J1*Ykj56BF5J^pLfdgWNGto`W~(g>Qm4=)F_Dc11M$1KKMIyR@#PRH
zA%|Z^@%M>To^%-H1+H+j`hv9m$FR}IhT~%M*y*Li8H!G{ndrdR7fP6WSm4Sx59iV3
zc;<UdWkH4>HE+&GnnD-um-gYjWdeq>REO9h8&rE6;M{EqYR*&Q;`BhK-%RJFwbS|H
zX$Cd+Rl%}X5vA+K;B(wKUV3v4>$R$}^?E$4l0V}9QU_jr(}K$eQv6Ufk^A>#v)OG4
zGEdEeOnm`+TcpGE$Rr%HAH-z$^@wQILH*_F=w;x-kO~`!W=`UP=T<P1%f+JG+O)E6
zgtB&j>~zm(>OL9n36#T)x*Wb-FM(|;1qfI&kt23EGq<r5az6ri?35IzcaWwJg5m$u
zhE76!cp!MY&aII~%RRwkmoMa+`cc$z6vd;)DtuTjcsBAJ86={GYmHa&Q|blo9Q1&3
z!%$8hI}~p|=g?KqqdX5+QsUELv|q}lPq!R>j_F~@Gi^Lbm<DInQr>u|%p?(Mu0N(h
z@xJ+t{5BX$!7li{CY)yywb^lUI;N-erb&G=^HSAWm@pKnSG@VlX8{hCyRg;l6DIUA
zrkB@nnh0?&f0qa!8f&5d-4xo~Pr{wxKKvKvkAAnUXukUdKDyYkYj+u9vV?PXr7y19
zR6?gglm694oO^5>2b~#<nJB>R$AXt=PY~<Zt;gAJRjm3u70U&W*;(;LD1N8JCGN%e
zeK(zP6;_-%>?lf(deJ`EmT%g$ktkr|H2o@gzH{KjvUu#-6wif9Nqkc5$^C{N44Nuv
zI299U_%ogsf)2RiNFoP|UB&&yz{lV$^wrbAM}Hx1oYumzd6N-;ToPk$3cescMIPIv
z%wxF|kQLdE{)%Z-wl2oKpY51vTtRue4EkJe#GsWTjQgjGm0^<1|C9pBi^t$>Dnm~J
z&yQ7#Wl%#BXRRn_@eUVu*W^+yqzyhU6*PUi2?w0+qM=HR((O|awd^ELM8+eztrJrl
z-09_#jhC-tX?QS~W5-MLaKKbnCI;b<kUQ*N=V6|Nso<yShtJ}5f^M<^bLK3B)`0Q!
zpBc&pibZ_)!j?8`B~gkZMz(igxxoZ3etQ$A&MDD8qAwrJ7xLI-Z!Qm2WxCu{TIA}Y
z)$<U1jRNpj&WRoq-=kbVQ{esNpt12FIwdW5GPNAmhWgC8IUeF|jmXMx#D+c+khZbr
z<cIo<`0IyXw>QClv;l6K_+m}RPSpAO0|tlD5NHQ8D=$i0RWR$K4q_0%KcZO-H#X$1
z+<IKQ7>M=`E4rt7V#lzvNZfn?U%%glYho*!G}iz=fqc9=nLh-tL`T&-goJw_P2UDC
zdB#-8JcErv7W_5VhhIkPLba<Kl6MBP?Rh_FnjC;aMkoH=jljXxD!A827k3K+Xg*ia
z-9L_Is_jk06e{xbHNk&-j%Z68!FR5LSK_T6n@2^XCqn^(6@*&)>#*zhUi^`&L|UO1
zj?a~aX7WIoSB_w(;}{;++$4C($8m9p1S~(PKr}@HGHOY1TB3ssr-SKlT!QG9EC%(-
z!G-W%m|67!UU%Q%*qJ~M{yT=thOEMzudmRfmCA4d)6dU!M%~P2oE3QmsRn6;SoqPU
zWE&cO^~ML^?ZO^0r2YpTWEcQjzPI3ls3v_r%tUH*Bs9!ovE#`OteN%+m7~V7Epr3b
zzE<PSUenqAU<J;WuEEK7u6QJ!h1MI%NN6&~S?3Rsb}{AaS=t;|s=}(uM7o5<Vclyx
zW;Dn!GH8(C1x#RF*?Sz8vPQ^`NcjIw=kU1!96M<V-28M=ZsLS~CjA+sJb}9IqqsIs
z9%?sdB3UDoCw0~%{EH6HeHZG_g-Y-;4Z@YN8PE>e2)DygRExHvtIT%nm^%W0&*^Z@
zD=`-LQm4}-eH=1aiN(j;gnkHVT1yOM^_W?B>Ld8d*H+-igFP4~a~^Hu^-<v`ON%9a
za8A4wb6S_9c%>qL-bmqx;i6bETn<sYz()mw_u{uTw#*ENY}RmI*lNbh_dK}1;~PXx
zz0kHx$YEaAtca|^z?gK3#Ry*1NWt5(d_BZ1Z22_)7!2w^V$h`i{I#wHJC>eC!5u}8
z-8P9T=405n$QrFC9gz5<hRcactd+3iq-zo^`8|*U>#P_tJrTdIsX*+K57gzzZud5P
zjp@Q3izm2vN~qso)!=FGAMh;DN4aJoABL6j>Y}X}GAEWt+RGS}tAWKM5-_D{J6;vJ
zGTUz-UgVEvjBgfK%$tS%wI`rEdm$3ejOChCS-6%b;%$XIl^>^YwXzYeG^ue})iK<8
z-w3ac8Qe195xTxsz{GScHS+qRM(7uiypb%li->d0&3MM18_Ndg1(+mi$pu@i1&sC;
zomQiG>Uu3=BbP!^*k^;hWcd?g_+YCc)w~6bDC!2P#^m7EtvT>@O62-8TVP`2g7Vkz
z@!@tFojURmeLtJQD(0B@UX*or!|>iui{qEfL3~gvdMwUi%<8Am5Te&pL6d*sQA`_U
zDP}*krTmjpW_WLae7guAN2>Ck<UmdfP2y)up?@I#AA0W~>rbm=#$ItM_@*IgW+p<$
z&V^GiMbyieLgbwv*L28na8?@auGz!tZ8YuG<6yD=4HiWVV=qHrE?K-Dsye|4IO4-6
zT7{I!mF2N9It*JdiT(Qs9;{1Rd~UP@-{ud6{AUvkRI%XMo;yO%Qx076i{NVP!M=LC
z1TI995?99ynx_{Xc5a5`t)+0)5&p|C2`=%BKw)qc<t}*glcfxLZSTRP6MgtDJA%ue
ze#S7pxwtI&l7HmKar6~Ewp%X6RbxHOkX4|f)DQ-DKE&|339J=7h9R>)z^E|{X{BX+
zI4qEEBQ%&<sn1FGrMX~53=ZdNVVB%Kd~;8t{7o~O=}r}V{Mq!H8;xEbZp?fC6^{b+
zF<Qrv1yiT6J<OHQj0ZwT$Dfj)Gubh}9%9?3!^ZA1b{)uIp3*WT8;wQX`A>M_5<rg(
zKW5ra;Fwfh)=obLtSjKoZhb268;(3dxAxccWXGnp*gYzb`d5^&<d_PTx*a&>{A!Gy
zJC++C41(j|2<$l|i`E|kHuo*zyL(zRnh83}_;S?H5qx-KD$hI!Vb|+sM6FBalQBJL
z8953o0!FjP*M^GMd|+tx3OOA&At@#cqbf1%IBLq=phak3lZ3euhHQLZ49)TFcw$aw
z^irWccA_$N1dR-21Z{8zPRaH`=|CUoA5vuUMLC{ryo)Hi7R>#597A3R_(b3cwRY&!
z$fjP<uX7kP)Q(9P=Oef!pDi}37@*Uid0iIpe)<y=q}2KS(pMB03VO+_m9QFDfUFG#
zRMeV@5B^uMvCIiceY|KRnMji}Q;@Gy##{ljN2jQwL1sMLTpLk+CLiAGC(|SCEAoO)
z!12HkC@meuOGfSJJN*G}ejms}r@e3v9?jr+Dg5ZL4WBdbz-)6FS{;L+S!u<z`Z$)9
znqs(x22NeoU}R(|<Q-o_@pl^S@AXE=ZVv`t?gb~I@9t`f78icnf!OH@kQ!BsQ4=ii
zY^Xh7)ueJ=d=w`1^unQKNi45D2<vJQY*raT6R8;Hd`#lIW<h({JBlYBm0?z)HU``u
z$1EqI7TGR%9i+9;9$-oX;W0q)tlFo=(&9`B#9kf7P`hHj-kZyNW<rn1`4QM8<-v)C
z3!#x4#sL4@xVFNKt+nZ7U@}uK_#))W0A9;lh4`zF(f{%Ut_)j_u#^?peLt8*CxZ9_
zvaG+EgE!)_td{MIb*(C_E|%w(tO6>mv%!5kQNC{)%4-%nxa4XATT4fL-2MfYwS$?_
zdo;v-5;=XBBtL7l;KArGNFJol`KK4brO^>1HdJE!KY#3YIED7F<KbSK!ogARV7)(%
zL82;LmzPM9)%~!fZ6YVmPo%PYB46mMGppAxf#20cenyMX<JO1v^A$KGX$mzh7eGT(
z9~us=&<ZtSPR?+M6!+ty1EX<NQ3Q?`WO;Rx91Na@qG#)QT%6~Hc`g=A_+rNGo>`o$
z8Oix3Ni+_g0sA%5ES?pDHwTL7{oNl!Hs|x|9YI%hy9TG=QX%)3Q^6-$z<ElnYAwU1
z<YIcA-41u1Q;6^%jzGB!I3^M=@YT6ISz<}Ux_r3Vmtwo0HGZB;;BVX6*lQAvL2A>5
zen@p3yxoVgtNP=w%RGeqs>9{w1@Lst;rgYU@pP{pd@YaR`x;?imql~J$d70|y%PcZ
zl3?9Sjq-vQFQU*KjlG6r>zlFoA@rbr>l8Sffzy$AG>iGR#wbb|3W@Ul*t0{EGd4SO
z`IBKh32_cg|BOqI1f6oWFW-ES#jFd4+#+6$aAQyWSZEHpIoh1rJ{kHaMDTvgJ-F>r
zq+FCM?|%IOiQXazxn#s!ltkvXhbR}WdzZ^{fs-otR||ab>32xqbpy98g+3Z_Jv@D$
z%ts5yv#;b(dKZQ9+mEqS=#|5RFIDkgRKShOc38DQ9%l9)6c;$xp$_p}J8K=jYEQ@h
zZ(95jm5-4@BJ34+1^+xXaX_&S)#fHFPuT$RmyI}cIgR;WhH!n*K!MM9<KvIN&}t*-
z8y>=*T%t|s5i_Cxs2s<>OR)dbTZrl{jnt8AQQ)YAM>U#!qtl8cq3<g@s~Ne5t1-pU
z3ZB;WP<bTObURELF<6v_LJpgAHIA1YG;ypmorUM5ap#d5G6dY(|BW|-O~xYA`5p?x
zI&fjFJXg=tr|aTgLLZw3by~Y{^p6t*7EH%c_e^d&Aqi=rkEkwkCw7T6Vu9}j=qwQB
zi>^MrTm;_EKaQrUOR@Td87TNcf~WRIqq8&zh`94kwldEBj$)A13TSF&@Nn}|oah+C
zZ;LZ%H%c4t6@|E#tjqJ+R{W==g@fHwAS-Zx?MZ`i^!*$x`F;z|`kqYP)0bbB8qjo0
zit=7(ptQymiq+LnU$7Ymx2UtyN(qC`Xk*g3Q&5%Bg<i!F_}bk>_vmtbXx;_c&E5<Z
zygknx?J#c9H0p)<bC%>%{FeU!#ihrvBhDH9hMJ@Fy8@g~8}P!}0-FC4xbWSI{BF|+
z9uL1EHEsve+NG&w{Tn`~N|0zDioru9FxsgZs;Yf>Xn`)qT@Qrj5^+?<htuHe4(LmX
z!DZqP%#4$!i=Y*HQ5<J{(kW#la9M3eyi_ZA;tq_!75PB^xFd^Gv(}(yq8WGj?!)90
zanzKVg2rcxxY{xvIr?&JHg|^8DjWV8sKZ&Fa%_L|9i|K2xMPAlkG)RB&`EBTyfaA%
zH^~&+tBu~$)o8sEigTThuvJ|bh2vkq!F4)c-pZx4;Se755c+1$5<3nGHEEEb%|9Q*
zfrnIBxIz!NY>JTU9)NwL+_>7_j8eT1K_+e}r5C61y_`MD#kHU{B!;bZ3o$~?4Rwz!
zpy83p`nN7r)i%Id=LcAD+zn;L+N>75Qp$giK)rY+R$mr}_0C-Ez995tUYy1OFMgqC
zz!Z#dHABUugK!@D9ZPF2!9OOB{imo3wZ9piE<VKKx?HHe41n)BDMS{BP;%HwM64Rb
zfv0PswdywFeg44U;aU9b70XO5QOqp+hJ>6b{8be}WBFf)MGI^Me}4k+hs81Z#b7>5
zbLXCyf+m_F^rtz`M*iNj5GnN%yyxnW&<eqUS*iS$mdeYP8tD2NjhNf1&~nhBrMD)l
z<b^(-7zNzEs)t8$g-A-WrnFHS&z36Ub#X3>l1H<~Jr)LAx{w?aL(L#1<aO<V!stwh
zIcdXYx2?c|rLnt77L`gRaNTtjUIUI}kvL%2JrqaQOYqq-4J?$AWuIr>aPuF4K3n_a
zmrWQWHvth_hY5Y?cd@20oKsE4Q9|b`JOpi_-m(EVGqO2%bQf&gYj8r@fU?$|h)mAH
zq1o}g6D7{h8MB~N=0-Ci_6+~&z|{?I{P|p&uBTMs-7<uyZj0c_OG^wY&*rA5*4#OH
z5ppHPnEm4a;pi-*vi!a#On0|*HzJ7i3<f2o0v1w|f(Qa4D2jlHh}eqV-K|&{KRYn7
zy8{)m72!Sa|C4L!Vy(xA`#xvyJu}yq!xo5{$WR&K_ne4pLIXUJ+u`3J(}!`Zicz;*
z26tCvVC?2Rtl2mgyM<qMgVzLhUoe)flY~AKavTQ2ch{DzO~dY8ac`6(l%g7OZ^>@_
zwYNf(m+<4LyK<$`Bm~A}V4Ha#wm<lapc@V}+VUNjPnV!6OqI^-bg|~D4J<p#_;J`+
z{tk-c-{6HfplQzfs{!0s8;sG3`!IV}IXq7qAimTPGj}QB#K#!Q9kGPkzf4YVD`n$!
zMSAA%#gjKBnEfM|-fL#SKKv|V&uRflQ?cr17&d=sz^0=HaFK{b-|oB6xcfF<G;D<B
z-d`}bNat6f!%1C=pkHB6jvbPR-7~ME#bh4Zoo?fRWB?BTr;5x4zfp5x8kFQG<KG(t
zeE+cvQ8PeObCE@OErq1+A$<HxjnD0cC%(jrJ41xeV6O-LJ!N6lbp}Mz4~u2>p}OEG
zR(Hvyam_&HZyd$J9pMZMnSpIfd*YO*CKi{GRS))}!JrR&d{5?%A;ws?ZxTwJZz27#
zI_$++xq8eVED03(<kV`kymhAA9A{h-oP{6t7sUOL!e}co+bUU)we|8GzC@Z!gidX}
z+zN*$8K7#S4tDRGg5z?I=pz$|tkP8eUE)gbvhn!vw+)~7rn6Q`i<2aJak)x=sK{hv
z)j>1nDQR=Ej~Z9fho-^lnEz3SjqN{hbY?QX&G16tRt<F8c<`*aTm4#;_~@1J1P{|@
znDA9U9U|_{%g2!S(u=AYUd$67d+o*|)>}@*>P`uE&zp~&S<XB%Wi{Skkidb2-|)`b
zf<G0axB6;EqZOa=V3Iy(o&sgRR%7+y7<}EA%7U669PqFStuZGcaWaCfH6y7oUmZ39
z(Y)Ilj&A4j*~8li4^~_fK36HqWvTJQ<~z85WjLmHTaB)9Iy5*^DDn%dkx`)rSM?AM
z?NNiro5VAnp~I@BUHGa-nrW*>V|sWPrL^?%t<ju2!)kCz#)w6uiqR-F1AB!(EnuB8
zrnidEGSrh7HYCDLXnr&2s-XP7Jf=I1qWRwzq+S5+1C=>t5cy|<77A9GqVHK}tW9`?
zP$y$Lbbo}Dzn+|?BJvB_uTgSd4jyy7S^s!9CY0{PyUDZhx&95d)w?1`%ywkzx})Xf
zC_3FXrkR;BHeV`(N23J3`NcDAPXUy#%|hk_X-ed5N3%~f_uq7<{S8G(YJ?+SvrJ?(
zLO9+}pD{}E9AcD>3Df*=!lw>BgidNVawLt<X4CSeDineXSrOfWt@6^8n{EN?d+PZ8
zMiZ<0<ZyAd3q#-RL}%n8*gYQ3=51qn_isHW25-h}e^oxb)(7`vl7&Wg9DhqzASrzy
zn)21@yU3Kb;=GntRuf*fbhezcMDm77EKlD9<$2dIXKX6}dvO%|`l%o}D2(fLrlHnb
z1#O|>*q$~JC;S(n=d&*OJ3X1XE_sMwVU1rOHz6%i%um0%^L_bX+&*rOy{^J1zd#nH
z$M>V@{v8Z@5J4SL8_rV=W=oPK+iOOPoUJ5kFYB>S#U=RLInh*W0#bgc!h7{-UOHP3
zonkSId1HmYRTikfCG;R!2^<>j%J7Lgoc!uD`j^G>dO`ypziUNuKWEw)4rS9fp`kP@
zvFuD1uB;Ms`EoU!?V=&(37HH`Fye-O<}lJTM(*KoF8kgOZ)#%r@SqL49tMB(Po?Ao
zJAQ3S;*UAz$ko^7m;iO?*BObkG@6$TZ{mS*J`=A6^54c27_ix%_e=)Ep!Xdlxv4|i
znz+(!B4SnDAnaIduDXL;%F4`qvH@MDMDx|psjzwOD$aF9%+-&jN9ZUl_48%#L{&U#
zE`q<GBA!O~p_cIMN2&jYpHc=-t(Aqsq{sLkHVM9$^F$7jxO_nYF{4dj^EHb<`hLQ0
zQ5XI9RFhAolX>cAAk^P0^75@vOpMu&vItZ7EpNum;Bx-H<Ho{4ZFqm(i?e?xpnIVV
zJa^0B+GH)HMTOJn-eQb&?oN}bs$6F%>bl}c79LxQ2L;KDy`sg+NrSPdGXOVkzJuxg
z>(J2IifH)*7}0zgD~$CK(PF`0=7Zqq(t$(w_91`l6!c#^kg<=ZLT&g1ObDq$vFs{@
zO|)QY+C(-qUWM_BdfXnm2Y;s?gMCCfR!xlK_TNUBuyqjP*N1b+@V~ef)sM4R_`rKQ
z8TPz8+~?}hXTKG6R}Nu!k;}StW-=-xC!>qG4O?}(GE(@bN^UPl-k&s{*dxun845yw
z=#JhSUt{D&IXrH-jU5*w+2yv-Gn<zo$?iOCLWa;XSIj@JN#S|XZFKQWV8gCNY<t>A
z_|+5;VXef{EAdpUNn)zndYt*IfTM<;*xccOS3TEZ#P>G*ay$cXHF>lxItry=S?cw0
zq3nwaOlqyc-r>D*-!z!h{>h+CMF|VK9YxuC7ly@!G2m4POnVMRo0KC)24&-{`4aSd
z=qkMQ8*rn$7gg@4aEYn}d!4c5%k>Vt^*k28k_R(N!W9z2FX{VA6XT3JAhY^4W{sVT
z<<}m=uqlj&T~u){*q3)xJ!s#eiZ%OJqP;-}cW&o$PnT4Fk2ZoxzT%OsAM;MdL8hxE
z%S6AV*Tt2;dMKmepdY6P-^ORBc-A>qA!dsVB?qYCz@;OI)%k(Y-w6yg>VmsxLg`)j
z3s>?K+0}0w^d9@c%D|fUeSaWmdo?0cbGYAWA<DP9;nT|Xc)ww&$dS}yUtv0r`j0}$
zvJKGgegI)f+r<5EhDFUP@c1y8r~60pjiwHz?S1*taVF%N6rteR;eWoXV429N;kaEk
zqNn#^f9rWrC=X-S4_R2O8HdU~s`za<2XZxn`Bd2*v0|2YusnkYYhQ}*!wd_DY9Ls3
z7>g|yVzcx>e9Q{s_v2pNnh?hGqebTNs4|`<sKE1{F^{khBffXXwIf+<Iy0D6D}uPC
zN|NC&;xpA|agdmgjMF=T%T?O!YNH6bBhJ`+qnv*g5_sTlCl<5{7DD)P*p80oaTf!Y
ziFw7227f}?iCw-(vS^(Zi~E~%O2B6{SX@K)LK76!T66dfHC)V!$GTPGu9Y!C{uX`g
zs>~Jn8hid4r^~YM<8e~<8eS}%gB(p2K3eq;O;^Xm^^-15dfkDJu{vry=i^^o3eQcD
z#4_Dms9L3j2dxLNTl8%`#yGJqZ5YqZR-*r9S%ivtM5wL?XK%H@(jVs$lkJJXoCQ#7
zCuXW7Kx=(>sMINmY>WoXB{hULCG?u}TI|_riuYT3prI`Rc5N}-Z6Wf<A`|>zk^<L?
z*|vo6RlA6q#%4`CAHT`wtfo4=ZYhC({|au{dK!278o^jr9qDsJxND6f?ypNg!Y&h@
z&N+;ehmIoti8~UbXW^P-Kcw6mL9^m~c8#<{QCJ|B${a()4|iypWOIm}JQi3iz+{ov
zG!&V}gj6R!X`CQ(4E{XVu^rV7UdW65fY!euoN-hRzCznboID&G8kgXl!Eux$kxLpK
z*{#!*PgIPUwax*L*SMl1xrpw*5%dWvK+oUp=qh{>y$!8#H;?F*-k<lIcfsdFG9*Mz
zvHag~OcBg6*N<=Dyt&ig;aV8DT+AMX&i`_eFP3Z<vxq<+Ho8wm>;(;ss9g>zRS6Ux
zRH3u_Gc<lk!gMW9*z7RC_R0~wn6d^RLtmnMr74GI4i$N|1F#Xl=O4chqv4|glMf1A
zwN;8?6XVfixhH>2GsdPlC3x2u$LNCos7Nj1mqGtv-Y}ez63RS%*$8$6<@sZ|4;KcN
zaj{tg{3mqB$n+rYuF?^H+{p-dXvMA3<uqNa2)|i6yf^wEI=uU_Zip>ia)RmdV;aVa
z{`Sm;_gLvY2mcxmW7oz+wmnecDp7ln*=zvM#lio}Sxyr3p=Dj?;Lp`QjGq^UaM6D%
zyp4i|wI7|kmT=VXV|dwH0?j)n!!_$GI#O@p+Li8DtmVVttXX(HR}~rZ!!bl!+-Z5k
zIWBNNZp@V8k>(9>Ow`89dCDlZUIVQrV@@#a^#5W!5Qm=ih0-lY1f9#|`M@%^1_iKx
zQ4uEWs6^KXrrf{kEcTj(<7HzeJ$?%f$19loT+*q(@H<9-PZC^`tr+uq3V!W~!Zh0w
z*w$*Ge|a;8SGqDub}XAiJUDZ!7VWe>@J!@VRpWlcrSBiK+zVo)pDIpO58{FYFYwAz
z9tG(;AXO4a?T&J+@OGfW?7lp*?<#KhH$hCf0y8)F=Pge^c$pu>Ee`{p)V8N(dKVNt
zTnl%t-PnFVgzI<Lq20?3{mu;K@}nxSY1ocv$t8%tuFaW7nS3EMi&MV0u`B%y_Ns`y
z?xa3w`_Bo3=1B9BX)MpW3V+_V49bt`^dD0n!vvKiCL1_l;qf87-D;1JYol=OX#oy(
znDOx(Pv-VhWU@dxeN<G%`cVhaxqTxvlwaVlwKOs^!x(#RK8|kM2KBW^Abm%8xF)Q{
zf&2ET6L-rPGY{n5y@C77vgmnLANyW~veY|`_TQ7?<@ylordcs)RS_$_BWU(p%)@O}
z7`8Kp&%AuFu_+1Ka-*p@)QitP3}x(LJ<7fRfF$wE?{l!hQHwZAu3Zkv&LH6xyn<UJ
zEa{(+$&0^dqw1LqPKdeDq;@$5)vUpC`A#TRXJK!RJsKBD<MR+VWUTvwJrS{pwvNKM
zl(R^`(2r9hoVc$}1~t*eTsc4zbw)Zo>~jax14rTh7!OWY$->o}o$&Z%D*EP7Y<~O*
z3RMQ^y-fxAlTGk3ay(C_&BhBqFF5U5i!oUnu{`58QX$FhR|DDc%nrKw|NK>Em&55?
zyZ<0@4yrBkpoZCHc&>57=WET#G0I|akS_x3&Eb0gD01eQvwXY*BOa<TL9l{sQ}<w^
z%T_F~*@(ysgZXJ!3MB(a((Q2y5B@p>opLE=IrYc=;rSToA;Sf~^Wm{zI6wVr$Bcv7
zlo9#AuN&%7d`a+W>^DGi&mdY{?T+oI|DlIQKT4dEq^znuM%49WiJ0lIS`Pl-eW~WY
z2P@iYFeG+8p0}7VeTV{lhR#M-su8bHg_4?+_^EGq8b=+0?(krS`pcl{X*W2YlVeDY
z=qvsjqI-8Wv~E_$B7I35*{qGL!F#duuoS-RnT6)=7hzcb44>Z%#^H`QP6+MG{ifwe
zJAM=<sy$ixO&QZ}ErHTkO<bHJi&Z|>d}1rK_g@>aM$rm~wyM+RgdDCMwBg0-PX8jc
zlhBOs!LG9$+1C9u0yH)taY`7AV&7nLTLcRg%Fu11G!NuF5IOi~Ff@pOo(5okdn<yx
z2jhobG1LV2XkPSRNJU@9AFB{t?J^DyU&PPf>y3qF`uG~yn|12xxbQ&@$%$65jjx03
z=V%7I+aZ3P24<a4$CwGq$QeHzm)cD@!d@MlXUL=ZY874{s6y?CUaa;M&vUgIj?Rxl
zm$_Tv-#m!H)>rUTPGm3^YVm6GILL+QiCL*TH*f5Pz)CY}HH0x@Wf)akMRw(#U<KUL
z<JsAgxSl@)=Z70I?q&+_pOnDaiwU9!QfGMSeT*+Q#q97zk&9AiuY=FA&Eydtnrq=|
z?L)-=v|wRshyM;mSv+Wwgx8CUxGvtmB~PwGT7ElL9RGzqZhnG!QHi@V1e0KgV8Ts}
zV&0u%CT18h*lZ=DMSY;UDv$RTRAF5}3MyXQhxrkpXjLxW|C)niIiE3WeWd7>DsV$&
zfSMfRX%ZU?Is0x<5zHZ(#Tn>GseqBlb!sh^WAwW5JbWcooLNh;th5rc`px(x@(wB|
zDzSUk3WVkl<Lsd!+zAujbryNeyZPKw;=qcBifrhc%Q0vBz`!bpZ--cM(`R?Ac-x2O
zD_!X@Lladiy5PKN8+dm<MkzXDnP6-_);o*^8nSS3?Z+S5lBk(li*EK|+_32sd{mC&
zkn8}ai>z(R9TlvZeH5;zy+t1*$>QKA_-+0j-4nE+^YA6+zb!<qwLIgWiaxzs4L@J>
zL0ZNyJQ~{-A$DKU@^c=>N(b=saxqtv?u3EQAa;&a=Qhbhm@}pp5yD@l)Ke9GPe)?n
zflyApEc#q=4t8tm%j!3|te5n|%ul&^(iD#oqL+CzUgR{tFTlK{T=w0qNYk7ioa8@{
zV@GSF{*wakol#_sIHTHqQfOap%Zj;bTrwn)DQbc>THcj&T%M!H1;OyVbP*?wZ0UbW
z<hT~fVz84l3KZmUU{(guPh{qQ8^A-m7Ae64sr{`tUEZrxdHOQAM<{TostL1JK0sxn
z8v=*tP`NCPI=^L6ZhH!LEn?Urt%cvsl_($Hm*q1?L#6FI)_qBXrmiHW-2H~Xj@8i4
zypKj>k!(<Wf|REN`EIuOPP-o$wctw3Jr|8-4?F#jj1wM^a&d1e9ffa;;49tUhJEsD
zP=7Fp^Do`U%U4~H?R*M)W<k(&>C1CPH{tqg8qx*7_3tT1)P)b`&uP}|uQU*~5#n83
z(-%kGqFFxMfQj#IP?X+<^R|y*gLqHJ><~3!>12G<kjLu(JP@m(2LE*f_*usv3blVR
zzF-8d)fL0IpO_7pIr820aQ0bTi<V^p)crdT|IVlicB{GI`PhgYTspm$81s{P1W(`W
zi`mznLs@9;DZjh&LvBB2&y=NOS|r<*^$_;g8aJwjGx&}c3s-zV!;fP)AiWwgn<m0o
zIf1<<^yHa^{utt~fCq=J0S^OtcAgF1j;lg_yDe2q&*8IIDo4e3VyjFHjVt%zs7)`}
zynBd?lPYjxNCGxVNWeNv776*i;3~M8S0C=d?d8+)Q{p)kTavM6XaG-jDkEiz1cRd%
z<9cl|0&ZLdcKYy<(7<n-gi}M)5&bTXLU!7Cw54l9J$DOwEVbvp5Aj%)Xoxif1!Lw;
zPi~3$2;Z&J^e(@LTe4lD*R6n`q$VJ4RbM)a*~s=&B9r6z1uOhg@OzFbWAfXurAhEd
z1g|e~<}bMS31nNaU@ZR1VvpnnaQV;!9jB+l+u;iq8IPvcumwP+6&Ed5LfgZ>sI5DL
zK(9E#t{j3LUC8&9eet=%0P|lMAW?WWx{2A9Rbw`{td+r;`q`Kp5CBD=Aj*G~LXSuC
z*eY9&CpJl3D|~`ECtkpL+j&eA46nm3GI*Gwf#nqnXem(V-q&wnTi=S-jP0<vB8zDb
zE}T10@I<ttp%o}nZtgpwe_0YTtzLLP&=9`nG3=(-6;0L=$a1s8oOiiQ68D<%k}%5s
z6Ksb`a>!K~h7QYG1Z`AfjgkS3{06XWtr;HPw1w4MZ*JV4i`M-Gd^k<qe=3hq)5VF#
z8ckTXAQKOY!<krIg0<I2aoeSNux;|gHi^ENXkd)klmU$Fl8U<VhhgtHj<4PIv9pf~
zzbIzX_We7Ij#`Gf(fw%lQ3WTp+_@!Ha7m*4gl|U~k`w1a<+dyCoQj9^emg!%nvD$)
zdqO747)k0j_^9ZFu+1{OH`$-p8ck{5avgKO_2uXhdRS5(M4f@L*ry^56Tz;NkhSL|
zQ)w)2FvgXg&M3Tb5B~a^*fgxe-=^PjcyCyMSiyz+U0084mt+q4F_dY7`_O3D8^3nv
zVz=E*%)KOo(!A?f(z+5)f7T)X&{y0EaYkc!yZ^By8G4mZ!0e#OsK4+ILnmy-mdYDQ
z=`Hx7dxWR3LJl@lg&ukI2qp_(<nM@uc<?m{BYNmVraBgL)N^=iu|K+e*2L~X+weZP
z8V6>sLY;06O!|$&+rkJuEsElM15LaYS>b?;Ss0k)O_k6vL`Uf0uDdJ`T$zpoQ3VW3
zlEEFRGD`mWgCSe%k>i!ZtMlb?xjF(y^6ori?utL#hO>w2Wo&tD%6Z9Yd@mee*9ZBt
zt4bMnZ??w9fGb%4MQ9J!`%(O4DuTys$GM{6oGtvlBZHP>QSMwg8F=IJk*?^eW6FRg
z2X-n-uw~IT_{qv}{l9<y(}#6~p7RS-BsJky**2W=j^@j~IvAB7h(U`zXf|LpPP@Ot
zm<yuLwAN<Dw9)tzsls*c0UX=47OTAf`a32Kg>|<mrdm$Hh}SFe;J?#2ZsWv9mC0<l
z8-+;6Yv@!lW4(($2F>`2Q85*ei&bYpxDpcPx1uID0oNymu)cU227M{yKCc1jU!a8A
zEBz?tT!f42%IN+;FeUEG;)`w|JwHpM_@FPH@>F?OM<2sOl(FzMpt<raihs92N^pBu
zO2}f}F-O#|FhoJ%d?Z*}@y|gE%BQ5$Z<9Rwco(r@*<P$P5Uj{Lf11rZE<DsycqN@r
z-LZ#o^29w9V;pUz-eQ-D1~VLFA-_Z!$KSg_&o`XGKB|xiaOOKvw^c5hg+eDSykD{q
z-vlGI`bMY!uOX6nIY=7kKbSI8(Ta8@dCYyH#tYL@1%FkCZx&8LrEehLyfVS;rc5dg
zii2aqcC0_8z{ch8Fh$Odhu;mueAz~9(-zs+v%%b+)Q*Qf^WgNX8EZbsVnk3Z{3jQ}
z)WaQdn>=_|^$EJVJi@Mr((LY1OyAjs!k=S<#OpdZzV0u~G*%(iGX=%%(unRcfy$%5
zV5n>%`;{ng_}l5II@^YkJMskILwIf?*5KGC8#qkYXX2B5<j)kma&fODy{LlhYtcvF
z5qzPRV3@QQ(x8hbV?Xxd=R^C&eXfCDJAs*%&+sQ!nkkRcP-s0Jt8eK;@^ln89k<|f
zht;?$7seZ#l<?$xr+=4qQcPAAd9XAIQu+tH*9+#(&2s$yR6rZum)L(-fmfn*Sunc?
zHdl;C&-sGa{KtVO2f1+f^P})qZuid&_2%-hkLaih#D|T-6X7E=%Mm>hRiugf;2j7k
zyN)8`tx#TP&j0i?`Q(@)t}mE~ZmqkZov{VFP2!ml{EyB5iR^y==a}_R4K91aXzvn>
z@mI&9TbS@5c1qLp@hnXH+nX~Smm^w**tk~AzkVh1@5=YMQMevah9XNeJ)09PhO+w3
z8k8v5aALwq^xx*qKb!Q>UelGmY}VrTHDg4F$8k~0P5gCNL+_H=cra87N4A>r*PKLD
zd>2_TiN!b~w3(aYE;o{P<+Yz*A=M(y%^ljb6W(aG6TM;aG>Ti}Y?u(B&3F@YI`5KT
z(SlSwdhNug|N25RTM>2#(wG{s6E2IvKAmTvGBufDuPhO~(Vk~!jAWV6JPq6oP`OWd
zDW+tI8C)K=cK5{PRf()$R);--?fxf63J<NfF(m6{Fs{vwYrd#riHR&VrVK=BU<LN~
zFk$3^Vel_`4(Sp7IcJ|U*R2|i-o`VaVb_A(50P9UwGbCJjiaWfzVLkb!pV)i<s8Nd
zcOzjpKaL-A1Pdg_TI6~ha9`yeI?w4cyk80?FIT6WY#hfat8!?DE@UT~;+eMv|9oD9
zzQ;mwT_=hM6Q9F4=QyN_jCnIEhGsvbaej3=e|Acu|BMVFlnD*?umuPI6MmP8x`=+^
zjbTIPVP!}qhO1<->x%^LxFC%Y7Lri7GyzMD+;PuT0$tJx{}vs@%r3&|*q(Ts*WrKG
z+lg!5x-)FeJ=_+4!QSu88EWav47W#!L4SO0ErVu`H23fL0Ymdy=>1!;RCa7ff1xv3
zFCK!kj$=4|qzBKOtwq!}H6TNoc`c?~+;R>-Hpn7-pFTJF|3PA!Gt{$)IxoTF4HG__
zUxJS`%aySbPP`MO4hg~1UGO=BpR9YJ^12CZRfh9<q$=!w*>J)W;jyL!x2{p*e33mK
zv|TVMk0tZ+DSIAM(}ZPDcRt&u%evHF9Putv@IGCk+6q2?=7SfFsr)@DfP11faJKpt
zoMnz+`*KB?ztYFA<9awdqZmtUEdZe}gspYPLv?>F+1Q(fPI6c#?v*WXMUJOiSDGhX
zg|g`R%;Rj)xN-`5S1!jE2PqsD-n{F5I{fFp=!f|VDTq{PLb}!nNd7&8Csu;Lt(M0f
z_H%`gv<M#(%HVE(5Br>K5Pl?_Crr9>?TTnNZj;5U)90~2GmoaN`q-Z=#Ybjdy!t_n
zvMY44W%FqiIEY+ur3s%auYif#3+(N^PV{Z7Fz>A{+YIZl-!}v@X>RB+7k!KTNZ9vI
z!zg2U?7u!0r^la$NGjp%gb+v!Oy<k?!l&pb4ZkC%*tgS(cOSUXPx&(<3WQc2b_i*^
zY!SJ%AL0!IkkCt)wnuy^tzydENAyr!xEdzWj*RJ1iT#_i@j>Dg(#KqcP8VtLT?8#+
zNAS*TI}YAyg_C=CBUQHv?l0tU;q6^?t&oJ_f{$n#8it(-F6g`ND2|`B;F`*CR#s?o
z`auach?;n{c*fOMoX4l_wv2vhiytC;KhpCxo+z7P)Gq_%W<_&v`d-*NT|{$>0!v?~
zQBvm~>NB)4-ESz}ee^N?(g>dWsLTW5r7+!M4Vl9UwBLRO>90Hx+PMe4pNexrWK?%d
z74Le42bwfoX)dS7t*sVV-2XVPI_mLhM+h!j)nZp6xeU|bJ@o}11bWlR+?<|m#Yp)z
zioFd-@vY5a?Emo&GMYW{Q|QUxruHRly3^-i6SR$X!ZSR9a;o9@e&jiBk99>*QwT1X
zj^NL-Z!mF)W~}8^T<Y0_DoNw;FF*=QE(GySR3+XS4};Wm;f<T=jn#V<aiC}_e!iTA
zC+EbhZm$%~N*uXB@a)fNEAmjsHVhMt=FU6&VZNpx)$D3;##04}gN^yTs}xj@xnbZv
zv8UitGH>0?;wIz%T(D|CUank$%J^Ft8L<tT(t-~#XgxNH|IfiQx=_ju!daoc=#KSf
zyP~K|@}>B*Z!)xg58<JY>L`kCf~=@Tby9?GIR2W@LuVj<NgsZWkjIgS^U)<rADfqp
z9O`R#Zcrz#&DKHnWlz`}DM3TjigjJ4<Cjvp;G@Z6m(y!_cOJo#I~8yYR^({${aJ=c
zqnqH?8p{%^j!WU!*B0o+Jw@8)$CxM6lb>%JQ2v?>S2Q@m^kx*3a|ZF=I2**dyo0-j
zF?`cQS-(^gTf2S6wTE%Ml39dF)-&*ABq8lki5?l9=-ni|3zG6!nyA4-*=*i&Gv%cu
zGr^t)a`$MmXTIR>-fY2vf&I{NL7IK~OTzNSCY%<&!|vyN5Gb9<-V<MA-G2#uqtuKs
zI{*CLcBZ1t{TH0)xieAprXvg_1aH}i;|{9v<K!lEEggvLS7UHi+~Yg64Z#zZ_}1*e
z12uK9`}!IG*+tSwN`gJBU9eR!&Q6b*D)ynQ#-v>jF!8V<7Psjj+h`OUdnrO`z&6Cr
z?Mmg7GI+AFgmJfJ_*`KV?j?fy|F+|`uMU6by+#ioWxCfTAYAy4+(P#wO~#nl3PX7Q
zaR_q^N8)z=YCKEJ=d=@A4D&5TVv7Oh7AfQTago=)D|mK8mLcF(FBn}IN!_)f)D+y?
zBlabNU2Tk3^-jE&5O;77XMU;KjF#2IcwmVt#(Wr!+&mR%cp7rt79FlNvWInGE*^P!
zamnBy^sbji<nkEoe$(L}GRu^Tiyxu*mOp0bB`{z3W2Z$JL9fmNV^_(cLqcRs_X}Q9
ze0Q2QiC%23GbMDQsVCTK8}=3;xA{G?T@GPldmdK4m*wxTx{MJzU)SWbSikZc3}nTb
zR~AIq0WNfU?v0-BHE`gME3_+wZ~L1vJ95wA^=UaeO%WWlTW{e#F&z~`qZ`q`lu@pt
zKCLyt^<G(A`feKfT}k258<A{I6@5lvSI9k(WT)4DO#dmxiN!ID3l3uSmwc{OPZIU5
z6?5Fh4B(8&Ic=GP(YI4L*5EWs=c;l;r8|C3)Z!PRjXvD=0wocG*(P{$$x=n!*ZT!x
zHES^7pBeh}8O8ydY-yty&K;jMATy&HXSDx9(OW~XEstVxtu`M#4&^+-CkpF`<hQ;~
zDC?<@ma`L)aaIo|gM&Cw{tnK)9***E+TuJvjm>orgzwgd|7GY=MNtmBV>B4IX$~HT
z^n_!qI}3M<8hFq?gk*O^pri1guTaCzA4x!jv&i=3!u|FNNFGt(7Q6mv8yU_~n*y+@
zzcq$L#WPoW9#)6m#Bh5*W*@A8<=5^Ue9M9(UR=j&O~C^Dm4QJniBLK?o)-@}^0tE$
zq+V#EQ1Um7%|B!Ib4%J!u|dmdUq~O^hbLOH)cyV)9;?@2vqU&{9W`ZqiQvXvl*QO9
z>TFvggRvjONV`?|vR?x$X3C>1QX73ly|6)e`K3Qe!YE)4R-fC2dsECf+47q3T?()4
z<gKXwtVqerUC<nqz|rzPd{XR;y0lp4%<Kifx=1#)jbpa90jiV3sV8ZKxnstoe)B>2
z?i0B&h4Uz%JPccVNuttA60=i7v2B+!V_r`X^K~Ptlq^SQ#b!9!zQ-HEi`Z#53|HRH
z#F#&lOuHb7@i%JFF+}hr5+V>8y%7b19sjF9gN<=A|Iatake3qB(ab}bm_=rM&_v>+
zI5ugzi``EnFi&v*A3pYheD!LqzhaEEr<a9h7lOajM&gdSD=i+G@q%d_ZQAqM@%k+;
zUlo2McPa6%`p`In{Ng5ybKRt{$VY?a)hehxyb_z2-bd~5VVonFyg%oNY<HB{J5(%k
z^V{U<bZ`qsi+T7@$7fh{t_z%m$D?~uB6JfZF!|_A9M8Qh7zw^OTsVsA*K$~}H;Q9-
zN8zYSr+@9hW@vS>#2(S>-gvEns8R{8KQHoTy6S>+>4S-X(nL?8Lub9~Q2r~2KRsMA
z^t3Up6BUrRLmx_Ggs06!3xAxv(R4w~LAOoAL#Ci|vNX!FoiT5&HsTj<#J}h(xat(k
z0dc1=W`HF%v|OR2z6)3HEAp&{8g?}bZ+rOyG`y2zc33Q&PgcRfx)}eyL~zgefr#lN
ze1l76I68M2oR7y~Z^Sb^HT{aYb`DteSqIDhsA1w0J8I-9;7{mXL~T&USQQ1Ri5%UB
zfe|=$z>)(+-p<bMA5MKqM~uh<osqeKIhV}1>Zv4;7<NP3Ty-eh$xym97P=V?m^%Iv
z7InFXRZ?q(PpA<#nHm^>B^}BscFZ1|$+n+!aX@J++!q~#LR%g_EeeH^>1-?+YLCkL
zpE&L#Az0_`@VPCCwW+JI*jSTauR1bpW4nK>v?Me{EfDo@3L;z;X<qstmRJvk#fC0e
zsJa-@g7w!jOOEz~EI7LVWH@STV4vSv=yteb@Jd;Pggz6qC>_=x9gFox=fL66X*7Rs
zMNvto|DRoj80Ga5zg;c3@Qoy|OcuU5!311;QJXmz4YBs97U$&2qq<oaw#y~)UT{T1
zhUsv=f-7rg7U7riU2MPYfv=fq;+cO2=OfmTIIM*O*^^;*QVm)0<2WmBIG>50q-W_g
zJRLU-iK5?(Y>cMiz&6w`(}m?GUEIy@j?mE$Vbh+*mL+an)vq0L?Fz75;sKArNAc%n
zG!h!Kc*9^W%AaL0qF*pwqt)pkb~ss&SdK{#CHVHxGAukl9@aS$2>dgYJw#2URV2ZZ
z&n|prtit*G_aWw-U<&gX-dUToV^}P5dusEq$!B~`h@$U2cbo{)LXY3_&}~fSr6KEu
z7e&-Tp#$L8(1Pkf2c&6Qa>kIsXzOPTkKJZ)Ne_kAK_l*K?8G8>4@7Ppfz)>wP$_ou
zj8@3!uG1CJw>M_y`H{GvV#s4jy{RVpX6*FB>zzu-$kRhVQTMmq%EjRSoVa4e4fxv&
zX3V%}81>VH7o>NiW|j&1)C6*qM;mI49%8+R6*IG*z|~Foy(aX=8tn}J{m+}d#omZz
ztqr)c>kK4aSBl?>KfNc|BCc8ZPt63A=5SYpuCT|<H}>>zw4wSDGt62y4Vx-8@!(lK
zrb;W|k<DGi`wWCwa^gR&a4Dvk`r)0L86^4!ah+g-p1gM-dwjL26+ZyZ4q2Rg@&d9l
zVmN-nH2ev$hxxr_aL$|x<C$`3)EotSTcJ%0&-1+LqOZsg;M9*L$SYZo8%hdT_e%#)
zZ&za3aZy*TUWp>}t}Oc?_{(DM8Q--PPU3EBOZkN4BZ9ZF5p29?%?N{aaH$vzxhLX`
zX?y{{ejWaDJLULKu<|0br11Ky1@27O#Wq8K{ur|hE1Si?MJTenw%8kY#gqy^GI{ub
zBWt(W(eBI-1WzsDfskuhc4)kqCrNQ-P7RKgsxthU9;&+7bMV^hsJD^ggp5q6KFDLM
zoGMeT_QFugojy-IaqGM%-`-Ue?2kfLrbtqt6nN**YM72{L#t7T|0v&$csbr16OVx#
z-n_@<2mky(O%%KKHs>R!IF_n@ZxCl04f{WZh_P@$R$?K>zLm$+q-+?u>$AI}2MbD9
zVY}eJJFWMGUd2E@GFk`8b0RCACuZnn#Rw5+#LCh}SPDN_UW*Gaud9H$t^_=~YT*r4
zvG(RXl(vff4i5*STZ}ljPW{2$JJzhUF`;yu1V^YZhDm_AnDu%HZ-+bXyAI<00(0g#
zG^0kd8?@7FIPi2LvjykwTcg+!@i$rc*gP1hy94r0{Wx8)0p&M(QMRanpH)k-BdidM
zk2|x(F&d6`WptZ9jnl4qvsZQo`{oHQ%8D|s={g1zX5B)e;TRgc6&mdYFZk%1^T)qS
zIOXpwc3c%v?vj`z3T@IqC<8;ER`O;}3D54yplkVbybj4>WG^M=eAULvt7dGQ{RxH+
zaeTC>2LlhOGEzH<r-seN#^watuGs)f{dwpW6o*W~!ZNw*#(M$7sTEX?#uiP%w^P6(
zKO-(z%BS3aLvi|uJ}>`qMp*S`ycG?^e8I@j(ktRTXLFnuGoV?b9^G)<k*7{2V9}8b
zb{l;VE9VVnwUrdk7=1u>gkZgWS&1Pb;v7npMc!s(PCwh73Lif}`>rBw-}t~o^l*($
zR$Ohh7&3P%DfLPCN#iQ%>}AC{O0V$xc{Tmt7USiwS{_(fNT2t$R4%P0Kc%tK*oFpS
z_aIXqg%^~g`=mf@+cKD0d&csu<~Xiv@kZ<5p)ehI1HGi?;kMA!`g;maj+jl(l+I-N
z@FOV76?@fgMN_g=_>{$7li<7-^vaW`?ADFgy!Z<mE-gjLuHkUM(VHiaEfD_D0G_s7
zDm4BauD10TUdV1J_bwLf^?p3z9L!k*4xxB#9NVe}aH_L1dcIJ^`?3tG)T(jvG6m{C
ziQ|*PHAs-p=Z`+ukyz~{I9TCaF1Uk_#g3)<d7;?a$BMy2MgCnz5vgy#Lv_?fkxg{L
zcST7IXi~w6)nj<fLDYFaW+L-HIgZQ9qMo-bx2&$f{RCCqzb?T+`6Zmwt(FRnCiLoE
zhAm?wsVKDghLW|Am|4cP1H=xr0zKYcJe)@}N8x;EJ#4GRuFV1EBGWjGGQ$GUeAAVU
zFH&*!XbmUqyN5votU3Ns293ODVBv<7xV&vLr=A-}m4KBfJ>Vm<%_Z1+>?GD)mS<b;
z2@G*M45eBn-Zyn%$4oc;i5A>}6o35ZQApWOB}@)*z-xpf)6Jh_kAyI~*IePfGKJal
zWJ-(}%7k`xytGJR+fz6AoI8%xJDRxD@&_;49I)a+6&h0wS(e&``A!F+v{lS!TvU<%
zXA<qM)L~;UH7FaD@Zhy}u?Ook`hH607=I0pl+@(Pfa}QGB+G;iV|Z~{G7^*e^RL5o
zyn1~Xi+^d;>8=_kd|QEW*E&!#-Uf?z`LIdpDT3~4Gm_(&(U^=*IVGVB$K%kBXK>4y
zf}3yW;>ee`XnQ%0>5siRCr6i3YYsvqNS*CzUohqPRAvj$nVMk3xJfOAr=k)2PS>DD
z^dPK@ol3VKt0B>xk89phl*u`VUF{Fy(l3!2K4<Y%b0dbWtfp~oPYz=+d;S)8_$#2!
zL+m#RbHzT(F*r9X5s!AdP->&_+6te>FQXOsvBL#%qy7j#r#atR0qt{3k!n3b^mO^C
zmlIs&qq<x=E{Xb?zP!|D9p;Zu<y_Mtf<b=*CE^~4JY0{e{atW)g81R}Dx5eam7cR2
z&@^ncczzb6^s1KFCvg?S#6nGJdrfZbsmoub<zk1?BW$^)!;KH}QS95Bw_K0o$0~E2
zG5G||5f`B5mIXIi7bd@}#yj6|82C-bmbE=NZuVdfcB|x?4U<_ori6LQ<;Yo)Ln*OC
z&*gO`t?mEApuGbiBYdZ^68A)=EfTHsPU1vI9)G$h(60FcqVBoCf4L#;6h<)mbOkqE
z`;EY`N}ir+#lGcBVLkB+E(>PUM9HpPQd7dR#ZQpqc?n0(IC1g)=XkPpv|tI7n-1AB
zd}k~_Js!gGn;t@OXaQBGS<qJEH0tyJAWLX7FE&In(lmuf?7Z3WUpD{Sa0l($`w*)U
z%5_TPa8c|yUGFK*=VuOl>)R74h4G9qTZf#v6EMOyp9Kd)F=KBov&y4sGtZKLww2=1
z)ceAZ7R{+Y1k2^0H3lTk#{;=9Fg|u4R+(O~sEg;JY<E1A%jA`_%P?=pB^16qj{UQ0
z5vDXBUP&=bG@s5i!!Qo1@@0i!j;*`<0N<@Wd1L20BsnA@VtH@u96E*4K@v=t>Vxr*
zirH?J#J7X3DQPf)jlLV8F8u7f8>Wa{Nlyf|52H?t9lhRr;IhJEEPZXlnV(bXY~juV
z;~37Z@Ws)Wlh`fyw#X602$sMvd@41=ll+r7KWjU74fLko9RtpOHWC*H72|wbCe+u=
z!X<IvP53<pHKS{Ad4mkCx+O98o*#Dy7V+hkL0D=Wj{|}qSL&O@1+F8pM9~KK?+k}m
zmO9Sbc#1!V@|wm{oC+4KI_Kvi8=-?EqPJ^ZtN^Xu-Z<lJiJQG9aM4r?UI^=lf|P;S
zo#@7W3o>xDEQza^$YATRAyh~Y<d_$k>~YnY277kl(3nFA{oI>T-4rnL;V`5qOy#DH
zlNokSgB9<qIoY&^F04bAX&5vo*YH12!F)b8K<v#C&+J%vCchDT97hO_yq^M_a|Khd
zR}h->$5GkxAqMUn%fR00;?A#T^_FN{9@vxfRunL9fikm&M*Bc8?>p5NVnmlo9J_8K
zio{N$PZvhwtzc^8HYDJ6QWyh;PIgjn5(9+a;pZMZq{T$?$@G4>xYq~PA<M91^9W3M
zuLq>{p!PE(RynrfY05I3_}K^PGsEb8dOkY-1oA+a^$2{>i#NPGp&{<N<o50y+N46|
zpO-{lzzVW^?a=l~g4^cHiucM4-}Uq`MsWHsPE^C^$_RcOoy#V*D(pBab}dKTh3^wB
z?(!=Y^A<Z~*L3>37CeR0p8hDQkHEvq4*%k;83-CHDZFN;Y%Q3=qo0J1R%^}@twJ2T
zGnE@R`|xa+T(n%b<=CxZY-uZJQm+VcUhhDjMK~(Dma|~sIZV?^V(=T0WgAt^^WDX+
zhW(Ryzt;eIZ7iZx)D*!xuw&{4ZA!ZK;q3bbochazEFFu4ebVf^szU7EkYPkpGI!M)
zqxhta;Mw+ro5Ncy5FEBQBIji`<0%?%iTzXsN<6T^nn#m85fR~rRhuTV{a+VmCr(GX
zX&3H4VuG*VL@#kFkxx(Mu)|%4y(<){Iow#Bb>3X`c@)EaWSRZp7FK2s<9z)Trn<ht
zo1*^gv`B-s(63kB90`fqgBanS$J_6ZV7;6*)=m=K_#5SdJ08VV(^R0}G9TX40vK*B
zG7~$k81-p4q7)@WZQ_Rdw-S)Log%VrUi>6!hNhiK+?hBE|3yt@fkP!fy{HiUxp1WP
z7tgt1fa|@l<j1!<{Qh?;bPh<fNpQHXwM}5qUU!<MPUFu31Nd!&5?6Q%{q#i@Z!DX@
zB{!V$N6C)04g+Z^6NG@#lX#+!1so0Dp+oF*%gR!q>y_L1JE#H{HnJSq<pJI|8S;sp
zF0L;cjl&1y#eC8dGTW+HHtem?O42D2)&pshUC}-`g!cmDAnR?+_5BTb(P}!rY2QI$
zkt%HJ^qJq{#Q~FlBXU!k*b)2^OLIn1ca{}<2%oi@ySvx}riYw{Y~D2$_2oSsoLZZT
zbD3U@=#mAgUm~;W6~oV)3NceIn10p@40KzI-7^eWt}OOiJZ{2{1%*)SEBc+2w@|lq
zC02Cp@Yh>(2Jb!p`m2S^htAxo)EHAqy*<~lNoO*Dh`POpPZh)TO6lD?g&+4%;bo!e
zr#zX)*0LzPu6&NgClmQaeF_IFJVuh&Iiy&XV`iE(r^Wq4YnvhKgZm1`PagO6s-fe=
z!^oB|V|j-e)C2OEc)gN;lgqhxmL$u!mhzhCImB)WV}n;D<lk+;!65}aU^yCHH+VoY
zrWf_Ey7Term3TYD5jn!+JEeXASD#w|xiu!pkZ2M!l@y-2@ep;{E3iPYL9QqpVuQqe
zoSke-i<@BYO?F&goWhJekp((niMzK(@sFb)x&+8Wt>Oxz8ak1@!k6)*A3!SZG#o^p
z`urQg!=Ls5PsD!G)@nyCzbub^vjm_2ULQ9190r|jy*cHHH(srN4)3oIaCx->+C=8s
zG&_P($A_am{5rlkdT>W)D?Z=tLwCVW)^HO%#K|KNu&J8uJ40FWFdA1Y^0_{=oYOL@
z_`5Th*%3iJvSK`&2KD1Vk%xI=JQog52ap@k2U)rgad4%0Xw#-LKlu>itxWk!JlFA;
zs%as1lIvBJGW)O@JgP)KdPfD4o-c$-#&CXV-GSV5#O)oA@!{qdWaO`f_Mo1EHCoCQ
zpLb((aXqY$SwiEDD)QbxMah|bzHiVKyC;c@2PScJcox+pqv5~T1PgZW#ivc(FlvMb
zS1C>6V9g|&w+%w?m0`GjI3Jc%C7EjW3=_k}yvS}K7u=D>*>&3JJwX>oMa^EZPM<n$
zQkWKS1HBzbVwjT`|K&=+tDI;*WIzvbE`J%}%_>oA{}PM|Yn{cowbq&1%VKa?VI7h?
zt@+}(37?$F;D?npY+e>e+x}BI(l41SwwJPMP7O;Qr|`<0tr*fgfbF%zm~*s08%&R3
z)gCuqJvW)VMqU=$S2ZUNu43j%Uw%S4Ummxi?afryxd{)u&H%>E>W}P`i5ReW9FMh*
zVB!$LO!2bg(BlDopWcu9uwnUz^^jVBNN{YeaduHI+5&Sa(@zwwf~lSJ*G%;I+mR@|
z0Va_X*vD`@7s^TT_=BO`Ut7+5+s4zQBLm&+gs-T>kyeJG3~TDnn$ZSuy1yF*cinmG
zqaxkidg1)XNqp6=28$g%ku)iuZK7sX5V@-5VkWq*uL){<3l7sk!A7i2!RbBi!uuix
zRl(Ihd!-vzy_UpHb$L$Ru@JMIM{#VZ43n(}GFv(XJx$~g5?#Zp+1lLiH<>F3WnrDj
zBh394%J5!;`A@f&cTmd{*9V|AESLYe48W`tQ!%QpTremP3;%2iKS=m;Ku^KZDvn?u
z^CUK;RYCgA2(FPy<iABzsc(E8t*b3C!J!uIP2Es)CYaqUL-{E~cvp|Ba=TJEKa5$1
z4-c(Dp=<2eK9<f~zM$^XIAq5+VodE==5G+Xfw}_?ydMic{Y^9visPLx{kiYB9;Kvg
zaj3wVr~Sm=HHbNAXEUb9Bp~K_Ic=f^yKQVc9ED!tu_OlHwm!t7Dmw-WFVUFlL@uoQ
zhDn;_auY9xjUp3wsH03??4?gXg3&rwymELwdJP-I7j}1{S2i6{{?D*q;UR1v52E6*
zF$`Oz4NqTfY>e$jPqpc&-sz2(g6T2v>n=neTaHu07c)i7S!OrP$LVUH|FN@QMAwO2
z^me?o^`VMyIhUTF!g}!;R(rgJ;+IV@c-M=sXFSIe&&j-Ub1WPDmcVqm8RLcLz)kE@
z=)XRmH!_6|Gp$zeqGY*t;}jY!@?ltI6<VfbU|g3nm@XWL_-U75)GtkNX9UOJ$`r9e
zJ9Rrbk)y<3@x^6Jup!t0Q&P>KtmTJq;ts16d-(Ram$Q1Q=tuYYV_|>|*F2twT&sL$
zxM(mpIv?Q$qV~zsfqKL)7%9ZU?ZFYu)kx;pw<SoH5c{rgOon?;1xlm{?uqRoq~|2i
zSLg@Qrz&wkTWDCDl`y!YH>LVr#ETkvv}=i)YgCf(&PsB(tzbe8oQIva1()}E473aS
zb9bs8R}9h+d6MloapeMJlcrE_(qX9C1F_r7`Br3~9q&4^kL+ls=*05n+gE}G(H{pq
z0@=|0CdR2%^QpbaT<#vpecvlt<x&E>0l#5-V?REQZ$;kQD%!`{(xWyJXZN~^Y_B)B
z4NiceL;^I1oW#k&!*MopA50zedE%7|%tyB)xMea72D-xD-GX16|Dk<y7#9UP@!uO`
z)HZ3*IC>nm3!Y5>B75}k(Bsf{O?vj0;Y-g#wDi%(ghM7+Q>`U@`8RMTSsPvV^<-Rw
z6~^W+hTTG6jP)?Vr8k0M>E@16^YU4i=F1KP7aHowV0Pbef?ZpR5b-{&xh)Snvk|;J
zYY-bV^<il%_GZP+gIt+5bPapbuz3|;=}4d>IunW^$p}glnfn4=^glJ62l`e)LVhRW
z&x;)F)6v){&h(%S#XPr6%=UbIx!+4m>^s~7%bi|am>@VQrkYe6DlaSzZ!qrfB(7g^
z53-V}JbPF$;2Q$btx)VCsF9&ThX!Xp(V+KRQA4!mvFm3Q&YV4zm8xP6b|8T5XQ$Bc
zm>+w&xU=e?6I#R$<GOZBR=r5)fM1O;x}iqS@aC6W75Fh+f_s<uq<z{{>=irK@5;;Z
z%WJ_$pH>9JAF90MBY24UtI#k14(3;fVNszrXmt!duZRqhY7va1LwGS&0n5Hj#P~!T
z^g5Zt3W;j08==Y`24{pet;iJnC#Zd?BJ!0-P%N1G3K!=h@w7H43BSI1jii{>A4Yqr
z1Qcz=-gbQ(!SeNncDy=*>credvw%IPzd`Kjd<=j12&+Y&BJ=Yi^xxZq#<`w=<P<(z
zy$Pc$^U1P6#)VDgso&l(eK4H=N6}eERk?0a7?B1Ar38@>M5IK@+H-<{L5P%qAV>)a
z28e`!q=A9mh27oV-QB3zb!^4vuKVwd;g7>$v%l|s*P8QrQYPv!3Z7W26^qA*D%kbs
zXb#=!f#){mJXSlJ*V77UAYHFRjlk--dno%dnwKSu`MjVn=cEszjrLl6K9a*Pf2yI{
zd@UBtbR+hQH&S>^ho?yA<v;*4<ogYA)T_=(Vz0rrkmCwNmfQ2=#2Tb5%9Wjb41a`n
z5|3Xs4kcbip^YoggnmP>MkDdnVmfk{%l~JQKhLI&MAP2AxOVdrDD~1t_&#Dw{x<Yo
zunwLn8N3}UfB!xWh-#*epFJbR-583#_by>ef%Mx-2U72fE7M*D(f?C5LJpPi`nXYa
zH2W!ge_^EDFyjG3W7yw}=De<f%-I>mft}rXy~Ke(D`F9_$bbV%bm_3FipTCp(^M}8
zxzF0M)7DXR>tIT?+v;?gt;Pk<&ZF3_H-d(YV?t&z|4mHcfa$_H@0f=rFGE?;J&{@Q
z8vKx61^dK69%^=9_{kycdB1`g;jNMVxhEd0PR7fZ(XbP@?sAJz(n}oet-SH4zcUO!
zb`{IkYCQ1W2|Ke|{EG<1f^V%^86U&4$FjnnW=P%XCOG0NbL2B2a55|tM%ggF=zRft
zhT2RR+mxN=H07IC-=WdY3y#`pEFWlqnuZ0K*?R(t{WW1-;*aXir}5^235S(U#q+`Q
z@MeDjHzk`Qe}@wi9L_-7`!+hfn}@lktMK)EA|rZEM8yn4wjFW;xlvu=u_1-4GiA@W
z-i}At*|Si*5*Hf&DmJL~<u+lYv=rx2;;P?@j&oK*`+6aY#zf%CmR>ykpuK#5%b9+5
z1(u8%Nw<yVd^5q5z8_t9Ax?TvKZ~KR;meXKR`lyyz(BK=sNF8l-*cIGsAqvaZQEkB
z#sX|UGY!U??Zut-57xQO_^#7mXzWi%SU@Qw77PD=oh-3+Dp|3}3mY_xaqv?qLnjA8
z?`l1MRr)YpWi+NNY{Ixb&G<5NIF<jjftCL{%(%1z6CNB!=@@xNPl!W^w=(N0cj8Y^
za*tmmCeQqW8(#)t>`@C?4%UX9krxc&x^wT$RxFC~Wbx5{^!L%=sQI3}VA2aYTfF(K
zL}p0V2KeDAJ;Zlv2rm*2UxW@WRH`CO-c8k=22pjv6EtxfEm-31_$1lP)&<v)wf&DG
zP|J|H^X1v2Xhh}1yRa^(jm$+1`AKF|WoHV+Yahak!q`(S{Q~<_6*Rj)kS?(uxmR2K
z=fCqg@x^-B#csu&nL}Z}L(Yf4o%y$y4>y||&~|Ab1P4VhIMkir?%N`LP6-lF%3~U7
zOf-(;1Fh9~ynxK>RLF1>aL?@__}E)qYU8W0NoEG=CyV%`y<{oV9^#sGj~d<FAstO8
z>@RwYAoZO{zTTgy+dXh^$3OVzl(3|w4_}Romvj7OTs{%TCP{^yI-(^)eZtwh^&5CU
zo`AA$kI_HL2Mb5`<Ch6}=x%!%&PNuZ={s@!4s^#Ebz@v_u*QdnpP=HZz{&K!{Qi9;
zS7~(=AEgn8pSy_8Gb{1(pyZj0!%(N+pg7w@g~OF>ajuQ3WHw8oRo{tU90u_6b#FR(
zF2L$ugW&n~7|cCopI<K7^`Y%J>Tek@Es%a$|AVMgk<WY6SZ<4sV87q}<UKFXt>jR6
zUoYkzFD1UIJBVYFG2DOMmG>>g37TY!P2JCnZ&sCWeTMN={V=}%R!F^R@pS4shL<vj
z;KwO{Hs4x8oz795ec?7-BpWs4P%}nK_#!ya60s^x#gQJ*d217CHmi)T!dYl$7R@8g
zY`C~xGKYSTWtl;Dj-RB;h{Vg-Dtz_B_lGm9i4i(kbjN`2m#|Cvu5*{Q;02laD<3=u
ztv(x(c0HHFoD_W3s~Npy{#M%TJ>FawkK!&pt}<1}@A<+UTB^j^!f9Pwu@391+VM*3
zsmLmh!}bzwe7)C?>EjGw=Us!mEio7$(vcJAH=_HvOV}Sfh^<cg^H+lS2JXH`tJo;+
z>|ep#jWb!-e<)kHbw)#MAm8~+g{k*AK3yQ+y}P3&*D3yyKyhim+=sp^)0kY_9>b$8
z;qt3DFWw8miDPo^TiBj1vj+1@FI9dmub}ZhGrpy8NEB}|y)cmb(UBRqoY?N%YJ{!c
zF3#&D7VFzUwfYjaM%3V;@B#mBi03H-FE$mQ(3D*9>CGI$(~}E0v8xMwE=pIr$0amd
z8b~&8j>>~Z_~b>}o(@NHpR=&D>`3kVN6|@Lk7@fPgLPXN^EJCraMweAhAg&g2*utt
z#$4xi6dOt{aiO;scY6&-*R*KrzSP9I?hW`i@E{zNo#g)fL40+T-g~WdJ<T}keQVZk
zxQU$D_BdLg&H3-eVcu4U?cKdtf2s`ETlZnygVtP-WCw@lny|gP3X>h0amd#0bWrhw
zh42}-`8sj@Lp`>gT8^hv<y}5bI!oT6xR)Qvo=YA;uebx|_4tMS>XCB3Er)gFcTAhT
z1doSoL`q2(uj-GbuVkL=k~YBWcDeYmd{E|>#I)pceoq;XYt@BpwMF>De{}e-vJG$d
zc#a-pviK%uG^d=DXzz?PPCxI?4+|?fymxQTt~!O>s(g+_3-*-vz2*JxI6SL63OiVH
zobwSZn$(u-{^&DVx0LGo5$tikH8!6RuD|bUWOuGb_03b*cycgp??vHgpDZHQRi1N0
z;m}A8Fa9oq&&#H4E4!yzFHJc2O*)X*o_*8&`PW`Lvj43XE|UtI>guy@n79f&CFiH4
z{C}PURUh1d*Y`X4_4}8i#bQ6kU2jU=FlC_pA)3f+r~6siDGg4*u4fvY;4A0kjio&2
z=M95jMF`h##<Nq$a(XA>t+aAyZKMr8)Z20K(|kVYQb4u-Wz?Fj!Aj4u*y-%evRhe*
z{g%hus@E{5{Xq7*70Ny?`*6p38my*;@t##K!_7xCNbf&5wMgP}^_#-d^Wys1<qWIy
zq0YT>{K=Y%gUMaxtZa;1kMpVdeJs}XzK&Pp<V@y$22r7J5b}2d{0=$e{D)37e!mP_
zcRSEEIgb0RyK&X$rg$>rFQR1*JnNx4XMQlJeGe79zuW`vQ@5dqt;|%C4`Q&cI}i5^
z$4oyXp7hc|mb^n9Ix2Yagc^F(8?)J*x7atYFGF57g7@r^sQvLu-i0&Kd*vTR!h-fl
z&@%#;H7H&$JcEogS?qhiD}93A;!R#Eo~`S~L9+(Z;BOZ?TjcYGhPWC2OoCR1bnFwx
zvZkyNtCx)9zucb~?c>JLuVZlVTrp=aSEl)oYz7Z;<%0IYaeHjfk>SF9=pRP++rq)s
z4aMkwg&dkv$hF&jc*&+U&WOh*WNQ+_Ua9l#`?VPGDiA^XuQ4mM9F^CLXr!WtZ&uNG
zp{0vuimo`fwGR6a48h1anLg~a<AM9hToN20XO8`tvgaD&oJY~M<S1BYh5x$h(za19
z-<BT63RNGBee21VN@2WvF^)B*!k~Jjj{%#{Nk^uXi7%7cX>K`=?(9f?nSIo+Z-i{g
zY)-$_6AK)hut0GMvmQ0!kvGzb{w#T{e>wQoDw_ZObif<gU*A{p!tWC%IC*I`&Ogdx
z)7RbE+Ru^unglcSk_RW$b!GDJ#rWE%Cv?IVVEJ0fFFjA-{0u`jm7bo$bR;y#RdSR|
zd;G8z57ZMoJT}c^!`!9}>u*BO#~o3==s4PhKEs8Rah#W9!AsNcp^M8PaoV<%XL}_-
zm{hR&{z8~|2;1t6Irq)U;&77;F1SKW?70PTaN+CAiQG5)B2wI9&~nxwE^##B{IkYv
z=X4wMYFw~vW*TD0PsWP6Brd&hA7(3eL2aK7Vl6Hr@5e+8SChTnUrjztm<(UTFb2+A
zi%V(3@fm*wuI|GaexNaey8B|CxtcI+Ti{)}6Ev*Vkw3C8jwJ^3RFX00PHBWzGHd>{
z-UDN|#?gCqEW4+6q2Iev!UU1d!h^G@Kll-$kDS=|y9c(Ex}f&f9XK57i;JWD`DSAP
z?BpG2m{dmX>z(j(X9e}-&S-n0ggK9*#l!jyu|qe&c3m}Qmq+m9H&uQZHkLstV@c~c
zG@Fzy_ii<Qb=(BqfI4iEeCV8%R>)lx&vzHEifgnB6*BiPeb<eJKh9y7aU%82yK=?R
zWLWQQ4zpBi7WyP(%bN$tHoYtkmnQhLNILW4zMkRz6_(C@Fy-|$O#U8+RZ+_k9NiNK
ze}TQa7^5sfmBk?iJmBw$iGTYuHnI{=5lNl+cEY8WvsTVM_(`X#emT+KM8W57Wo#+@
z?%)35WE0_MA2vs;3F6RBH(;W0AbqFE?rL9Sw$V6_(ZA!Mls}M<ZVloP;j)`ZwrQ@9
zHJfX>^Y%F%T00V2UdLhgE{GXT7U1)z6Sy$b9ph)OMtsv?7<AC0_1kzlyBM;`+7`St
z<u|fs=g{k+2ZkLE;LnpqxbYy24g03zXT%!h53!@pe?!T&{pI_25-;xE$B$!~!ld7g
zIn#gQ%9GR5GnYAcdOn+1l+!0t824lTA~p@;PMd}NjVe@j3dH=oi*e{*3ZFW^M_qq6
zCion~F?lw}%HBJ9>l##S-UoMWH%|Cv%A(j7D0#XZI~IL{j_}=|tWHE#_ytsM%V3tN
zbei-s7;@7A-^1LwcBc{FpC7^hf=e;7!kvMssxVrofq!dKxp{LWeRc9-JyPyoW%;x=
z(vlo(1}~+X;l@)H>>bxavgF{bORec`GFUR+6EIlY8@KYq@WMb73wGJV&R4;c2}T^L
zxQP|AKl?As5NlffQe0C0r_eVkLrssdOwfztvyI|x8DLDm_TyptegS^9TYx_K-Z(V8
z9iOKy14FX8M3KSerY)uK;RvnH^7mFP^DL7$P&+%6L!T~z?*KWs<ZVQe<S=jCAtMIc
z;8FJ__<SBr&UlJ{PkJ!4(@0D{G@2v&gi~K>6nEN+Cves(yl{MkG35c$k4@x*lEd)x
zzKbCT1KDMuuJBKzcx&HT<jS78vtu3$78v4qe1Dcme({*IBQx?zjh7|(aB7jb_3oqJ
z%Y6PfZX~_zT4Vp}a;A+C_hRo}IQykHpB)S3{=Zs8;27B%E4aKYoqjFGp{Cvw6Q=#c
z+RW7`o7@-;&Z%hoX(?(3htcO=E5@iCfd0Yxa6P;Y-Ydg7w(=h)<+P;oV^!dw75l#L
zi|GYz_~~5{eua3!yE!<|NH_rl<!@l%H>61iX2pkYtnN6HL2HY+;h?aLH|Fqg%Q$3b
z|G*`yau#N)GQCrtWRD9tZ+$r@-p=LmHZ|Zv1-uL_xoL44=W6xi%po3lIMx9AqcWiU
z$AYi=T5|fGc*$IJhpTN9`V6x`!mb4TjsAr6m%;EHGa7^E8nLz7Kw@4iya;N-nfJ%z
zO7&ot$={N0gY2lXOZfOpb0}*?V(12Ou?;ss$EL$z?<s7yX^EVx@6UExuAG^sjp&9b
z1iM<{;<Q{&s}yd%bVxI#pByqeg-0`$=^8QwlQ*iP#neA=NvVU|tOjxQ?nChlE#&GB
z<PxPVn7K~6tA<Lr9T&oq-k;$(NPK9r|N9yuXXM_J@180S_!Fs^Gy5i<=ju>>a4K&N
z>(7NTG1xIOk>j%3<J%0f$NZt_pe}j%tPJisCVy|aCcHIiBtMR~$B9RV*kjO%+jZlx
zp+$Sy4==!oLp>Q7{1X9zi*WIn8(;Zn;{47zxHaS|ke$nu<0|+uC70Pb+VqzD^_Ha-
zjLCa|uzLv{-mE{`{tke4v_4<$2*xGlNpS3@%~`lD{FnBU|1cE)t2gKMmAO5W*uP0J
z7Y*;h>6O9U6{L+5Z?t5O{0M*g#bCnL&U`(%H~%#;K-!otG*@y&zT`%erpa9NRvDT*
zbwGD>XV$DXq}CPj_Z|I%0aZ2FQR%`VG4Ejd=mjc|NQQHR7eCLHT}bEs=peq!13m$8
zYg>exh(t^|`dd-FwhNa?ey__1eN3)B0Mq3;SgxTfGm?HRiOT0MopL6AQDUvQxa>Ep
z)A8|k?EBM$>E(9xQ5lI74_iY&z5=mVgRyY!c6d#)U=QJb-tPJkXAYRa@RvK^+!@2(
zzjoo4jsrS}j%LJFJMN0{l^&rlPc*xPs)>s*Snmd&YeWOpa_`m&!q#bbF!yUm*lnMO
zhr$HfeM>TqpTH-j%FL?Wj?dX^Flw_luI?&ib{BVOK%JJ-z5J=t4*pxjeGv)pMG=+;
zhVaRjG-2H(^VoV5d@p~DVFUjmc3vE-Yjb$I;UE%Ce`D#*{xk>^55e0oPAqwixx0i>
zcku&W98}?tgb2pfr$}!@97_7q1yT791Mhf7q~&3Qrvlxy%<(l;2VM$w>0&oS8?8oI
zl>ppPQ-%398@897{<4Yx!CHEdM@+@{o;pObah2k9>xpfRQW-p=To|#oXf*9V^t><)
zZjyUc?2Cj=<qP~WZOeFZkIWxq4v+b57=9y#TEbA=C|TbpG81%t+LSi}&cWwK6vGdl
z#6r_hm^KFdwubW5cuy+&ZIKS<bd)MtGHRO+%MbZultlzPMK{F*B|FSk?~Eru7Qs|@
zdfUVo?w;9+Rn3%n;g%U5or>eA@b1*@c>ybL2{*1<MVQXP(Cj_|PN(W{EMP50Z=Z`Q
z`P{EG+JLvW!Vu`$8~W$&V1bK1R>n9{#ZQL`XH8KtKM0wDE?iNQOhf7GcDl3<Z~QvL
zV4W|$E;Pe+$xiIMW{--LS!nmBGdIk;1HU%`_@>uGIGm~c7+Z{&BORm%tPYbd<mb@N
zNH@MJoc5oJS6{Ta!8@2wODeh1sxcQn>dz+ZMzdWqSFx7arQDs*pX`qJNl_>Zh==)~
zU+8uthYyz;VQN_=Yy+buM>ZbW4T<cz*b41dH{!|L7VH!-2YH`-P;<YW(FH@9-Rdlq
zvznpp&oWk=8pNkQsx;f?LMOQ&sCzjh`b8-#vX&#QUw3lMYXrD>qV#D?7TA?>dTV>e
zJN4kwA{Tx>+!ed?ixHI|E<^3R2zs*wX<kmO^ibtt<0+_}bsd)0rg&A8j-iiRBWdMw
zv_2;OzG_31weBh0>BiV(<iu6l;?gqE#;QGi5t8DMk|+~yOLt();7)W~+nhVxJjKzf
zkMLHmEG&(Ku4*-&9tNAPIEhKcO1N-lJ&p!;W2&&ibgeaTtfB!oh2t`^L4}Dc*5cj+
z4>;yJ(bme7^^@}$wIGjSp;2hM#v1XFo>X37%iY33`}=$hl{ekO+`y9<v0j)VjVGZ~
z$x{6GsHEe-B0LP<h<5Ax^5VE9n7J^7yB;O8*T!OI{r3=>o>r*3kt5l{JZ{^z1hG}(
zEPoKesz37nZzZ!FGi!ctq|Tb3&3LWgBZiM?$zMsM#7{1qsq7ml+tZRIU48KFS{JU6
zEOzHjZrJ^77qaw(&vDloyE+`iE!PNswSNP>I$PLGG{K3w40so`rkR|HYUSMC>zEDg
zKTd$2WHrm?#xwGe3kNjq3;WD)ESl|v{`=SBYhgboRQ2Yo_aiab%#G1sE~3+%*H}0p
z7M4p~_@I6iW7i$Pfl0p=J+@vJf7UIW?7AGiJ&)q{TumNdE&Tu=6Z8`g+t*vReDu39
zCk4j{D|HT(T9XOR>74zi9hdwL;bE({Sg~gmHhk{Fb5}aSp{*I8X-&hGYzthFyv=`8
zyR%|QS8SOT1|_{N?5;D0AwQgP@!1(zHn<^s)iPv0Ql+0TUSi*c(#5U~NB)STL0t<}
zOTWo-iawMg3(-9K6$<ZEBh@^ciLE4?F#HwH2p=%8<9FO^VU6s>$p~C5UDyOo3{~ol
znd!s0;-tK%n#-MNq5?KmU~Zo%9`g)fU{_V){ODkgaQAK&G{(kiaY@~kKBkR1_f8vy
zRrYVO_|{Xrmz=NN>O8*8aY3t>Q!wbpWhBqnWBlJ2Q0Xb}wjJ}~oF(^t*-?BP62sU{
zh5Ru~_&9N`Xuo7Ue8a<dVS6TCPkD%_vDZ)=7=vq;DNO1&1xec$qHu;j7s)wrUU)~8
z9-AU_x)2uM(B`c}Ny0R~hU=9D9C1Hf?gbV+_A8sKu6ISF>F*G=pgWfB>4%3i(wS0w
z4D;t3BIs!yw#|v>GLvgq5I2I+-g<abp3a?a$M7lJhFJ~vD4Xwyq3LPjH;Lu&{mGnt
zMEbPBX0(b{p=ptwfX>a3`b9b(<(AUTxR11MFEOVwMY>^!WY07Yx&zfPs@Q|K-ig=U
z&wwU-;+gYvD56zO5V^w?@u8jh{h1}>55L5ZcK30`TK4j*mSghH_c-6*1u<RQ(`3X@
zKKsues`*n;9NmHkhV?|Fla?H^w-s+*kc`}hR7SXmuvyz`;jWJ0)GLiK=;IuGKVZyj
z7KI!=^byAIOXT^}(qZV<1cn<VgM3=vOC9=h`(4>ZZ?lBs2}|y@3F0pgM_ih^91n$$
zGi=5vj(cyyj$<krTV+DK*LqxWYcx~J{1Kle9PcKxgl*vgz2th?$?0QH$7Z~oUJc9R
z1GrDo7Jl-KSp0DSkKC`pi`xURyj_1<^wp(Zm@7Iqm}1PuI_Q{XaCUHO4m#h7UVF>X
ztLQJ*JXe?R#e8g@-k8gb`(XLw_qg8vG4uv%QE$*?)J(~xbH^Np6xU<yb=ePJJAu^!
z_SDwjj;x=3`6D${@=cHN_iqo@WY}|)^no{jsK@Gz6g1!A!&5yC`1p!G&7V|ZePc_y
zKR1<p^Ah-9tiquq;$qAn4UIMJxJBO%%8gB6Ex*^@1*5UyVh4IR6^@UcHBZg46}NI%
zjz6{^f1YdN_nn1U{#Eu)c~5aga=lfX{W*NjLCNdSf%=cV=(oFwTQm%yA$$mrcRjgd
z;2;bcn!<q6kGR@eop<-AbKpTIsx55GN#e3Is2zf<Enec>1uG`Zw~^1rlHN0qW5l|)
zY(AzhC!F_W_v4-nH@Jg7j*-GVnU57Sb@`x~6`Zx_z<xMk`|Bm9?)K%&fvchV`VWkC
z9N28)9&B+p;qAO!<T@Ji!p6qXYHJ6xMip%Lp&j#Q3SZ}3H|}_D#zrlB(LBc-zW(CR
zIlCEiQ-Y9QqsKtSD0Zuxg%nJOXPY#r?+K*E#rJTv`GL6cwwN|?7Sz|ChrO0JzCSUd
z<_=HXyWW@U=Y5u5Zy0U*+HwAb&p7JY9J6$#x1aVDk2+{lsn&r_4}3=DsB)}W(^(ks
z7T6S(g5rL1b}!1nvVrBiP+rD)&x2T;?!>XaOYzWSC{IY&tG}LPUL(tClU{)N-J9^n
zhb(N2Y0j^`LfQJ{T=e^7heP>^y!G%5KEBA|seBJymh)Dy@EV3(OQU-39t@q-lNG1i
zV_~r~|2!Pd*HbQFX}~7fA03XFCv>^f&W-`ub8+pP?2=lym5y)_wp~4q!_O_aY;7j8
zm)g*;vk5A1I3Y^45pQhBpvXixFIt0p0*7&;!vnZ!nKJk33?!y&iMvU`){EM($K`W4
zb>S(7ez3ygk;)u=UJY#{Lf9{8G$NLIA}6UqQ5e#o7(TTVCaLH0jiElTe0JfWOKl~?
z_!}ClgwMWyG;X>Fuulhf%sJJIdmJppN7)KZjFsq=Y0Ar&O}WGOC46tWuwd&643Km1
z`%44aWpg%Lyy=H!bsg}0R1!-v%b^mT&bPTE>15r7z1>E#`1Dk4w2eh|4_El5hoh~q
z5O=k4ht9|QSR}ic@r~!;kWCunef)W7m^p_WmQHHZ*1{%p=8F-Naj&Em@~*2fb-59W
zx=%vgXa_d<CBpWM3zv_N;#~Xo?9*M7KVo~cFejA{#C=n5I~(RnJ*fXpTX+x7n7pPZ
zK1VD<_co*Wyp8Y#-W<idpT{xajyE^_>x9l*FF{EcT-R$RcA46-f9qx(`Mn)<#Z?)$
zsR3B#ftdq^+gUXgasTyV=#Fx%pWl@)dpdJ^rW@(g7Q6lb!SjqEqPsUn-Zf)pcG7}w
zzGUoUCt=C^p9-h<&9N<G5POyZD@=Cd;}rvhUVjDGDqARuI>^tG%_~lA<Y)VfX%Y8u
zBfkeH2-jmvZ5$Ji&WBlWSB!o43eOT!@Zq;9lef1Pmu7#eZH$L){a?Ik;fse?b<wb9
zEE0?Quq<J=^yS3=^UVwAoh!I?T0Sq`u|ZdFZTfGpWW<cloVU9bgS-+cI&J<gID&uo
zAHZhSarjOCgx8Jaa~Ul?rS{h_^*-?FvoEI%lzxO=Bon&(a?nRp7-vfMU|tN(4Aw&L
zYyj8aQm4({5ROyVh3S-ToZmx(i>!yUgTDj27HYFr&K$}PqjBqZ8Cr}{pig)K9i@Y6
z<TC|X%ME$+iWlm(DKYM=<iJ+D;>wgY7!za3m-!i7%?$c{uEZC(amlj)uHI4&Gs*0&
z`qP)cF4Zd@)mh_ppg3f{2utDlbetGv%?WdJ_%mfF=MPN5GI7McU%eYuQ;qS&(txwx
zOAq}_3KQ*I*zm14T(+)|4z&-Tt2JgwrOZ%QG)LWozP!=12|tClWDu%h>OO=Ywsga(
zEvEe1-w&q3?XG+4LG9|7=#Zhu2`|s!)|gWmo+pmfNeiIU&zo5;W1thP!0e4{5m{R+
zc|9A3Z;hf_b_o5y>ad@x^nDUyxU5E*v&@^&#o-66ZVzIFwH?}PrQnkOXFT8UiVWX$
z8mnsIeD557ly}=Ib;<qw)JB8!(u+M*STxL%uJtDT`R@v>+bLnyDtRW=NM7&OCoES{
zmNV03ahFH2y;nbRggS}q;<sX7eJAX_`~cnFzW~RyMD{Qxj!rp?{DB!faL=51=aLwC
z;0(?!vZUtCQt8x}F{Jbo{5M&kP%;qLzR9y^`w)Is*Wkj?3Vgee&YC+lSkQSi*Gz1I
z@?dwmY`%oH<9=XkNoVvc5>8Uo2Mm|~!=ep~F!UaH$fFthpSI@S(gEzy(t;n)c0uO7
zJ}iIa!7TS|ez-OtPra1cy-|>`u)E>Fo^6;qMw@jH<Jip1kYP)P2xrNX%OlERzIh7*
zY}<>kT8X<1<xKs#HHX~UgY#`HdHY9>bm7-x<y7%P8NGs9yKZ>wdmkR#P4RxGHZ*jm
zVC!xLBAz!!m1_y2M?A;etwBg$n8~xhg*!0G5PMUf3NPJ>OP=?qo#haiEpLR%Zo@Ht
znjT}vERgx?RV@3`4>uk+W9PUy#K_rTR+Bk+-KP(l>xA$}^OG=s>CJIl^ik64hr(R?
zOU+x&MyCbw+`KLaK}Oza*JS|=rar*KiVO_u+*=$hx$IXn8%<1Y_&ul#B5PuhvicEr
zv`WC3@$GRhS`WXIgBUfY3xcBxm|-%SYu}Vo?Gc2rEDqf}o%zu%fuEN<A$VAGjBQbe
zu$=)oyh?Ufmg2GQ8$_33TD*1S4yF#Rg2r9R3vHT=$*(Lp@1-{9o_-A9FJ9DmJRbIP
ze$5=wjkAwhq3a)O{t=(e)?Rk_J@zzQOxx4b%L;D424GKOEo|p4#N!GxB$)&uqR~5~
z9#O*MEy~E$5k}R=cKkiJDc+>Wer(!HI8}G2V~Q}BhDM05d>mdSnB$~He-58fCG5RP
z(E07f*ZTkPykQKc9cqchiSm6sSb!6D1L^fNf$kCNFh1?bmz`p<#x{f%)&phto6dwJ
zZOOj(<@kIrd430S#mL>LtuDoiNiC7u$O*H9o*^S+1m_5gxS?HJdS5sL^XT4m{OZd2
z{T-m%#D_+PQ_(}t8V9rkG3$#CyPLN}(?Qv+cQ)dZT}SY+%!gBoUL)2&fS-Pzfb%rj
z87%F|jqYvu=65rA-RwcDzjL8&eFILNGdV4KEUr9L!?WYU0g!AHcVy7x3Art>2zwq$
zj{4epEZ-WyVCSKX+^B}am)Ur>Q~XFZ?eOcd4d0ZUfNRz+ERi06h{Xu5l>0_qd>8hr
zH9*#2d%oCw49iLy6vbcW;Kj7I)KpWKy@lKtsvS6Hs0q9eJwcy8gJf6Y0skRw@JZG#
zmCX;~%GNAs$Q@()P|00&_@%Ird&kRdgRrlAI2@N4v7~7l7YtS62m@WDu6+!X@41{4
z+lyRTh2OS=sPiC=|8|KV>=F3ysSYY9lrZ=3AzTtKK=3^s%>H~9=0X1S_@_bVeIDr9
zvyhYTcEgU{9fZ*$U8DJR@N*Bqzy3)aP$PH0j*;|{-o|j-bJ!;HaGh40@y;k2VM=p_
z2WLTh$w)M`y?{F1T9odHW^I+sST_aJ`bJZ>32w_kPZz2y4aEF}6>|P7M+=9c)Uvk1
zje|PaF+lv1ZtCdzZV<wQTC>}E6LH<?;qdDpuxt~A8{X2hst^{5<crQMv|;Zw$!{)m
z<E0fYEYT>YW{n!&HIK)`uo4bBGLUQKowverGwL7g#eex4(BInu!w-ntAh#3xcby4S
z>FG@<yNccMgJBo>4^JN5z+3&Zcw3yze?KCSSZ~ii&!XAywkB%=9k|<cFx@`H!^-#v
z3L=`}Z)g#pTX)0fWrcVWRgG=eW4WaHaDE@@fQD%mv~v85M0q~Vg&xj1_k;f1iI^H$
zjYikaSn1XZcFmh`+5081axQ1?y7|HcHpZ0;Mg09o`aDN{xGzEYXdI1{efoH<HWn=&
z2VqmIAa=-hr@`b7%<!3r=E9U-<1|1VL~Sr#=M<8bcEFVJ9oZpL6=n{>c<|B)S{LnE
zHNqHU+t<q6{~0>|Qo`>fZTMvZ85_h|Q~pnp+(nrWR+{70e-XH0>xpW~D39ycl;J7%
zbaLMU+gE#0rmBZJB_n1o+KixXb$HT}EPq}JrMa&7INTWQ6T-Kzh5sI}i_%YZm?qt)
z$8AO-XANk2$Q7o?J8;i~Jn?z7!lDksko#B8TB9;5SyZy8^tJ=iv$$+PF;jYHqV2vA
zw(HQ82jom<I&mNl4hZMF$79fEl^5>@n6XpnD7wggvBma!90?hZH?MnOMUoXdrIq7^
z<pWr*^Ou=|2aH?pLYG-G9|=|Aznd<2v)z#o(E_jPRMDgQGUol#XVjTCR6i!2|JW|v
z{_+KCPk+J0j3ip@k$rl|PJDFB#kZU=u6*1WzO$UzanERsPgUZerrzkeBARFH-4Sr!
zg?W{|d7@=gK8x6bgT^YjSMUU*#4oS4IGXpy8o|mjiw;2pps^vA?$>@P)>$by_PP!p
zbZd({{m&qxXEJ>iAJ8Q`9%-W+31fN>qBFbjV*h*i`$l?MnZhC;S&4|mX~=v$jF<QR
z5{`cl_?UEJSkrV4pWFm`((8&kk;>50A+UNNxtWXE>{z=Qj?xu~ojDd~_7$Lwnk~<}
ze#NE16Y*|R2PPNi(I>4QYi+&wb#Vf=49S4YbzSs)7eRl`a@0>R;JdgkG<$puI&wd<
zd}W5=2`OA7f2#{B+GB!OCI=lh#t=7e{#u>IFZ+Y2FFXg&6n9+Pu8IoBE4Wg!1FuJF
zpk2Ej@JbZVyD)3RcI-#$2|eMc^`Xfa9r23w$MRSUnlD)k&-_QyvG(H&!?rZN9>6=9
z6ERaig*ylSRSf!5iPO@}&>RuNsuwn>u#tV2W;a%)?1JSn$(yO~Kqs%U_@Xia?IrJZ
z%_*ASVv}*}_#ef#g;n_TL}rJxekvwS9DzkiU9f(V2hP4ff+xZpZ6A`y$$5YA;>KC{
z_D*Kl;+ybvG-A~rZxj>-u#0ON{#`KOiMU}%x>^X=Z!@vT^0vIw&9QLWXYph7MzDe0
zsnl)g<Yz<m@KUD!BNZB*Xd@h|Q@X<$+Ic9S%IsK6QG_i=gbyrU?U{4R7!xxIek121
zW1TL{o>_3>oSwA)9n47w8gMOtIA5d=#*zF_=oG7hv9AX*ab{;!tZ2ZH7#+z#Wnu;k
z>Gwf=toOCJ_nhQTELwBx`mLB8Jph%u<FIUHG=FJy<g?%PC<-fss;U}Nx6DBIlO53b
zyD<h`u*YXZ*{{YlLj6?vZc7irVS*#7zN_Kf-#O@Ck-~eGVen9x<J6d+ijMz&p>J#P
zFLn_Iik~j#?bt0&Y8N!kKZrQnU~D=#3ag*=<ynJNKKAs)YQ=Cyi$7q&y8=$XQNX`-
zMO-sBg@c8G`K;F<u4owurTgL-RO^Kx>2_J|(c=e4;d1mZ;t4M;6b_WAZioR?pC$2u
zt=u=$ld1XRGDb+I;Op6zc(6N{J+D+_(jse)8r=&fNm`hrB(CqRmNH}LB#gYV;=?n+
z%OMjnr&~A^Z0BKoZ4$>9<*?vSG7I-t;P{0woSc}z@fIpr=uwTC7EL72FaU19=gA)0
z28qH>zgpWLM<;f|EA0T8w|KC@VFD6%M|1xK`8O+7c5Bv+buIoXzFpVkD5E~;G*E@-
zvfa3)?g*~dABXbX%lN0&63tR__~C~#*6O5keXb*0lv{Ji+W}OxDMs*>qp<te5998N
z)A#vR_?>tL_ffm>YuR4hEnJL_nVp&OqA65%_~GQo<9M5uB)y1cT+_BGGql>Ekwz>A
z9yF)toI(^&@51;GwwOI~Jqp~6@mOOb{(i`%_Jv2#tJcHJ3!{Z$+Krvg#iQD~B}11F
z#h3*x`F4gk9Rq_|EWO*QA^rFvJRH7PKI7c%-OzWGbLq1bXdT;vrJwrp-R{-6kf#C-
z3t|38ctZPa1ot$$h<;-_uv*;(PutiFb0r<aZLOgbKb$MQRM}PTcrI5-(>-#gvAKrK
z3KO2G8HeQDLHzbOjBjTfvG?cyP&wlril^Cf^0>j6P~<7uiWH_+XV9}<2z?Sd^H$fs
z+_7^t0v!yPwETyni#+2ZRt<tdjVj$=>GQz&rnoJvhI8wGDz0qWg^z=yv7_HOB=uZ@
z2Yq(I=ASVg-kigc(SL=nmccd8U0~@n4PT#1FT!d($`5wMjb_4los@%X#wj>eVvgK?
zZX90YkJNx74m<2d#T6es9HWa}c^Trv>&~oGONE)-1oe;nXtJg~jLc`?-J(JyoCUT>
z-^F*^00!+=<@r)|$$A*WX^}do|CSz(%Sv4E7v|71Js4SbMxy@|<od@mGp7-iuDpUn
zsX3P$x5I%QGtqi#Z?<<`gJGF9*za=*IRPEGt|1OjbuVLIa1nE-*mK2uJ8`m<ONJ|f
z`_6bUyv`6EM;9`8xNyX*J9EJ9#_(R$j&}}PqH1|Rtj!B#X5L>#(=RrB^7b%-^`hA2
zfetO>?Kv~W922|G!%190jcyjp@-$IUGE}_4ZP9y26jO(XaqtA;B_DL)c<q%qANK_{
zo5WFeFqP{zl;NCa59#x^f$o4r-dh`qL5kLN^lytklS9~eTPCkQv*);0<rudzo_<sN
z@_B<9$KRG6X+i~FrI#@7{$>0aHULLUtoeIa1R6GIA@bZ$tbSV~J}4Ko7XIV}LpiVK
zsMDw4bhP|Zj^<B;_;k4r!krwcso9y^QcH2LTL*MF0mf*B<Kc~_!oC~I^<$<(_hLNu
zcJ$_Gl_>}>aNy3|M$i$qUD0z({^_HQ%W?;K(!~b>BUKqJ3=<nwRqojw!kgh1yc|~!
zg)5n5+=`)nQc%0!O74h-Sm%FG`~ypI_-;QesBg*W9p;$tV96OlSK%|>1Gb&xuq>z-
zzIfIv+%K#`p+hYi4>#m}$t>En5U=OdXu5Z9kNaZ_;Gyb>b3eq<(#(`L6~>s{!jw;&
z@5BQu11!A{j#V@u)zBS-rZ_NVL>zy5mD2L=Jh*NBj4{e#TzvTuyklOY@u7P7Mdt9=
z*lJvnJ<MX&J+QI1<MtVmoM0io^J7+wxoD4t<8Q*WsU|x97>&#(fLF_ElxCa4D$)eo
z{`BHWJ56N$@q^pic6g*zg_fZMnDVO!L+&P{dO#tksouqVC1KKJbw#~&$@+d=1j7ei
z8Kxr+SDA(VHInDvYa4z`e*=vs@tC;3KSq6Q!}qH?!uh8jPS*$W$M0d#cX4H#X1%!1
zPV$L?b5J>UBMO}&IB=6Mmuyd!44gCjN#CVn@GnKLKttgvszPsR6MS}8#><>G+;boh
zf8}2J^G8ee()Q!GKORVIehXDTW=t?pmf7?enEFT`RrraAci)Ff*gbszmM>k9cwUJ+
zh?}wdu&S_~?BlM%(`NwRdiZfpUo9H!*oxE%t)-{>0VBHi<a^UBPE?m{QJW4hK6MHM
z;;p!GPAMO5YtO~g)e#fZ6p@|t=sU7IJ#>YSu}~Y|+*)we{pNiA$qR2zC-clGE#8zE
zsitI|4EsbNXW>Ge(^!ZT3;S^SuTqrV_rN5L{W$a^kD8a-bGDBq!`4N?_2W4NhqmL=
zjI#(kc?9h^4oii9cUP@Jv3I4+fa`LxW%MQKguTSJ^OJ?CdlEKI$1tI*INdf`@V`MW
z%$WWMrom(IIb#6loc;romHKcOXNLKACz$B?!>8YOB-nW1lu{n@(vr9;L;SkK2IF<j
zaySHykmtw|JYD?-itr-o^bf|WaBY0_dWjO5dB1OY3PCa>I@h}|({`xyX{%NU@U&+j
zZsGMM*^4wNQ%}d5KWd9{TV*Pox3s{&f-DY<n}xyyrn37JPv*tWXuV(fF3$~RKhlq#
zg>Uz>AdxFGTj0k(AFTQ8!J0>15SaW6`?D>jL%#&Kbz^XQx(BxAtVKf)4-QcG<&XNq
zNR8CNxv{Rogt~xx%C}Mda1vZ5yhNN^1;Q>oM2-6`Sb7IzsboP?CojbeVd$0pXpE01
z*I>mPZOLc_<6#>UXuDKl%gRJntY}2jXN6EY?#}o?WzMj+;mB2IU{KT+E-M~lqv?Bu
zjWWafmE&aQ=O!LFBfN3$!MJ0Opy*kzm{<Hm5z(tXlQs*(_;NBw{g=%IZ7a#@jKjzw
zY1~#SyX@!L`1;Qq&s64PeOwD_?COnsH-2K8eCO6R(ZR5&9$XwMxt8omo>(#qm)f*I
z;ZNa1jc~`ujKO?ZYYHbbM~;8&Oykfp=(J5nfTkYa?>Cnj>KK^;)(N-BL7X|0u|6`B
zJ>rAebB7;Q<IQ<8EQDTTqp-?52P>wlvoti3k*}{HVNf0RKCQys0^t^zyu-&gtvD=M
zo{ityqM@P>Z?<oRnA=zJxm*b!g!R|RZUvfLl78Tw2F0Nz8hBRLnR&JCsk~H0Tsdj1
zA2y7OJFdci?tc^)3q4udUK2LYmZ8~bPd>}+O6S_k81T(sGIfnOy-^dsS~(iqi_8!?
zMSf48dO?424<6`o2P2+@qo}Q|aB8a&c6BEPDc{4N$EMi5Q96fO!p8ZT0QXv9Kx$vZ
z_63=AvzPNg%wVWp^k=z>KNY&Ji0CW4eDNa>d^``&GF;ihW;6UX2Xaro4%f7iZsK=m
zJn5*<H*-wcGHNOkS6|2N)&9KcuL`ffx_mAy*g6$U-mCA+$hI0-TK*s2{~iey$!@qf
zYa^s?2Aod3N2kSum?S=I6=5l>N#5gz<w%%GCcH?cA5Y{(LG^qzho%o?r)}cFUE7yA
z!&UI*<UqJCQ^pMM82AaJtDAOfv@O+Vv`rF6^;hHS9g+>mYKQ-NCh^ufnL7^%W=rv?
zOgD4H&kc!qHLtBO#ZRMCy(=z{H{y2LkDZfTPRKhqBrk5x>5}m(x9NcTTO-&zum$`s
zMzM$EXq2!)G2?~{l#egNmluPvL3;TuU00&uo<7?@QAdM$y<+w!HFR`L;_!KWSkQSl
z!cT3)lqVkIXX=lkzE<4w(u0?F=W~~KF0$X=#msYgY`i*&L$^4>#Mg)|64eoGl!4Qi
z)5R5c5Ay@WnW&*bwa>x{Rw?0zeX9JpSDhA#U0|Kp1n*AY$BRotxLr+L!)EF@vP4`!
zEqvIqM)vM^``}2S_ycB^W5X)R6+iUii-}`UEPePtTOYtoa$o3q79Gb6Pb07+Rdyai
zwKy_*Z`9<$(YjFHCf$R%8H_!XjVhzgID9OOn}4a(aDxh(ix1|G<Sh>yuY<G2F6jSh
zjTSH5xkSYja$MyoJr^dm{i)bz6orO*OZH5;jcYr4(BbrY%v);5E_*|9IY!)gd!jID
zTW3yO(~La_Hs<NS!qtA>m0#8C70LTGFn-TdywTRgN8xQo*SI24H-_dNHVeO2lLM|7
zBXOPFZ_hQS;iI8EbZRY1A3j5D?karxC`_9u4_Zy0iYBFgwDB$B)R`49%kZad-C;Ob
zxp9taIJZolk7#FOULEikX}gjcxnwVv$^Nr`vjfuhC~@0oZFH17N&X*izT4BB!d&94
zYGG?_7PodQJ*c;kxlo`Z_p~{ORIeBoe3tprK_%vUYa;t$BQ_mnk4TyOxeTs=XWA5u
z^-bnt+r~6_z8DSH6WQ#=8k~H$4!K`qS*p4ef0pP_x%>oXRa)}aaSK*mPKD*ltq6>7
z4s%Ooy!YtO<7=kFwnPJKXLjY5tA-e}DFFTKZo;FuIa>QjzvW2|D%Zco8cPTM$iIok
zs@90wlfowz@|-o1e97v)IPqGzyM+drR^N-xmQ8SDes><t*Tn6nOK@d+OMGsw$rbiZ
zSoF_>-<|})@AW-gKhz45C#K?TVhtWe=5pqn98T7mkHE<)Sk%r3Uo$4+w}Uu|{J-Kt
z(->ZNw#L?^j$CrQ3Hr8Z!^3jFdiFd5i*H0?#ML&WUL+$MD`TNbJ1REch4b5R`d<s;
zE?rN|-MI-{yFbTG`Rv}yb7t{l4PncBv$l39I|*ANj?>V%KUsW=NH89RtQA^#a`F?#
zPCEtP)vnwW^#|ksgwf>eAlW@{MvO{Z?)dssVXSo#1^0CkFlaa(Qodr-ubr@u%;%nq
zvoK}o4@I6uz2eNRmK+@3ii^rkp!LR!Pj!_gi$6~|;r&<@CyeMjPw-TBb#q&5a#gl2
zTn+o<8rovl>;eX;Y=xhnn)vRTqs0_k7R|LmZ7b<nPOpS-jT$xk&&94u%`i7K1k<0Z
zaiV=1qh()}AE1wB#hN^F(~2*XwYf@WiIcY6Mo5#j2o3S%-Q{LjJ@^b<TYto_)Rk~G
zw&L9O!Z7TRhpn1J8Qszd?ef>4t=0i#E0Y$JRZ)=NkDu@QVb8hUcrm62|Nd@^FPqvy
z*<~KOjFIzX=xH=57jKJm61(*!1~0sc{rAdXf^S$*dH_THG_mILB512D!n9cF1g-l8
ztHmkeVK8P%ag^jdM+zHq5MLA;WABw5%(#=ptco0AL_ETXPcrimU(eIlr(nDLyW-c3
zBz!y~cLcK`i2h)QEANar>6$vKoV}rusEk9eYT$NUm7^9bp?%6X#lN}$^fk^y`Mo#j
z;NlOx7&%`JJ_fgvI|%oyfL|{awEx))7cK2ML%#_}CqIIrwjmvEWU^&mC5E1Kfa`{E
z4DBkub?@eU)n4AUPCYqlYc|gI*eXn`-B5lQf#ttCvPN>R?%Ksvduj#s+{R4n*q)jT
zwqf@dWi-`qjIFVzc#zhas~g?Lt`7mUzqJgJuT)sI+7#a>Tj2VHW?VR`JB@muN2`v#
zIV&)S%Wlh=W2Ze%rP>KU-;Jj$fDz3!dBRENY@Qp@<Cq$XF52;UUM;fYt?>8qO}H2c
zkJGm+pGt02ZIUDXYJMn&jhl*bU*gbeqY`G!_YtO#?E1RK@{&b4e)-ia^sTH}W!ajI
zeN&+~d=gq@reodW!*GcppMBEBahKnU(94%_VZj-g#UF%!urDJAHfOIkgLv5eI9hf~
zr^y>(``XMvb!{D@%x~jPjtN(ruMvK*4z>s8!|6X2Y-y%~lAXc$DILkQDdhc^1Mt`S
zHfH5n2^YJR_^X43+Kt84I2Pw-48!%%4D??TEX=NhaCLN~YHT@`EH!X+c@Q67U4dxt
zEL_QunTMJy{X-Sv^mOIm*EM*O*^75GEHG|h3H0tRf%gPon&;=iAaey;D8zXnnWHW-
z>1_T!ic41X6sISo)8j{j9%gv@X(cun#BiqaHXP8E%>A_p8kgR~RQd1pH<C}#e1`XL
zL#g;0hw{@+gf(Ku2CWjbt7^fEDt62{<HyI7r3d>UfYnB-k`WGK`27^b)rIp(Srg=`
z97mkFJvuyU4)4jP+#d7`ops%C^Sc#;B!|^Qa<~a&HPNL*Km0vo$iz#n8K<%Uj$4fw
zaeg=(9bbu*<YwqmSB5;LREAv}gSUrlS)9}uniXaoA?})k-y>Q3AVPXu$=F<%%dX`~
z>@?q!`{#Yboz#~2bSM~uLxk6Uq!j(uHpZiUwP=(*2ffc#;-Ya=%sn?9sRw%FORWh9
zG}1zv>Kr5{>2OAHDBPC>;^2NYELZgqhSze8KiLHr9MTze)rX6_#j<Y83h<^TOtLdr
z6>=NtehT*9uFBq1eQ6!D49WZdDhv-T#C(f%@uay?Z`^eJ{#eMMIX*O3-j2iTUtv`T
zRSvzaj~A`O7&O-zH)qX->x#}~%0i5&QsT^utvGL}J9UibV^f?Lbq@W9yRSQ9*FZyI
zN2_7~$Zl+HP_I~1M`jNSXL;|}C{GcN%Us#nT{(jn55Ee3U>FU<_uz$ii4|*iq2;Er
zcw<t9GncdYd4U#uH&($cWsBU|o%!)?d#p<g5&q5{JP7c|H;-ZlWC^pPG>@A*s-t47
zJM9jIGI8f66s;6@clI!xEjDLJUsc9CCZSqb1g+8y@SyoHZrk9EtS#+eWV92(U!Or?
z6NpB;`}60yKwke(i#ma)pfqRzGrGw0%(@9~_V|n^kHx(tU5NqZ?kxD%2P3Q;(YAab
zFa4R1|GtdJ;=w)`_`W4VG>4&QN9F(XgIRm05iE4NbDV{ARJR!M_2Q0fGGscG-`YTt
zY7O%gAI_8CP4^L!gZJyt!^Lt(mv_#@NB70aqsJ)|-^qJ=rDS{>ki4g_oQsUG*y#z{
zeJ{tlC_~tLr1Sa@M`%9O=Eeo;$ZF8Xsbe}&o+=EW`-3?7-y#H_7{<yLPoTA-H~eNO
zbI)jVOuzC2fz!iqH)kqdm`mR8g+8}S#(TJ5OZrXd!U6iSUw+$-KXk4l>>as(>2DNO
z8Nha>fpFPxqo_gVHBq&2Zm+`)2GM+|BK<`BJhr{R9(rGnBV3#d(PQjjm3Tt9I*s7D
zq&JI4s3CfC5JQq$pz3H-u6z6h)8>Y;-6<t@|N0G+UMzz74ihwJOU@@%fk}2tF?!1=
zbU!M2oZ1w&DE*;$P!UZvBXLf?2&em!K`763WJ#_(I~S~i&%Cwr8F@%YYCH1B8{pKZ
z33xSHnUysyIpm2hUaAUXdB<<09@{7!sp+^96U1Am0@!@;SH=9lk8ts44Q&6#pr`Q1
zi}xELsk|v)ZJmIE*l)tp&F0ky#`M~#f+rVK8QwvODjnzIf!18CNo~*W+7HFkaTZ7G
zn{(de>A0Dx^goWyJFdt6ZR4e(p|lf8CDBlchJ24Bg_M-GGD2h|Dy1^gl$I44*<@vt
zz4s<=TlU^7E9*Ib&%a*xNPWAm&*wak<NcO>M<Vl58{zVeXmoD%35m0w;bCA)j3}zZ
z$bYi0IJOky_qyZ7myw)g9Lg`>)$vL4Pq!{B@#J;MgkKnq2<a=NrjMc1dT|_jx?}l@
zHmurs4O@ns#vN-<oSNj0v4$qHtICx#s2loux5OoJd|LVRfR*`mjFf!pTeB;8ePjdb
zgll!o=%*rfZeIr4TXByt!Sj2Gi~5-d^XB|e?44rG5{m}KgOD6#X4%o-_!vCL9mT!;
zO#D17j?zO@;c#&x7WeImj(g+SZNWdqgbRVxjSA-3o857<hc?f>4uO3_Jja%eN5l3e
zIB8|f#n%(jZ>A~EZE4QdRT;dlZ6rCpVLV)HjalpB==jQ>I+h(!_fUFS+m$iFT4we)
znsevufs9jb$*z8rF|N^9tg5qS-_J{h!)d@K&9B2@jUCHpSo5OH*^GRCLUB)x>of*)
z+P)r=r5wm7yY69roGmMtMDSGbH*`vIpr_Mjc)VKxt<6f9Dm-mNFWGg}{=u@L{TTQ3
zAUvOchP{Rv1FCjHO}HXwH!erN@Ik`n4S^>XquNihH?_(f)v*s}ZndQS&8B=iIUPMW
zHxU+bN4D$V2bUKezzO%kaA;$~3dIPzFA*n(yz{4|{(+`b0M}-yAXR;o@Dkhcg{cnv
zeiSZdMJ(J7#ju%&8EV(oE36l`hyH$<Bj<~U&pHOPoi&iKuq_=HFT@{DYrLK1${|-;
zuzzI)%fiJ4BA=bTa@I-CRYCdmq1<Xbg)^>1GE?StOa2pH2j@taARYZO#c`lJ9x;O*
z*z3CNd=7eGMuRG@nx*hfUp20MB%Tx3shD*wn%b5oTu|!@yLFcA)zAp8Lt?pn<9wXy
zsfPJ8t)X5n`KW?$`u(WFx&2RYZq#-xtJCJSXm5ON<IU`@78tW43)@t-Vr;5#i%%Ac
z+v=C%gFF-0b&=f3o^Cu)UIv3%3h`9`Q8-8H^XBb)*uKmQx-%OTIWF?E3Dcpw|1?b9
z?aDyqMtJQp5;eyZOpM)tB}3n#cd$0bYj;4bS2x<a-hllHM}BtGM)klSiWQdPdFq{k
zMX#l+wrvK!Uy4ScvoeQh_2q>=s`xwClRdr8;#uFbXtI4U$I83liq1H$`8}DLjfbOq
zTt{wt+K07y^|;elW(JBu=%dyc_fl2yE4c^-Q$5i#>M5*WYcM@F6v?$3IMdHayeuks
z)6IvSmpb58=~Br%%e}6<f%Lk(sWvqk>PkA$Hf_W%Go-^9F_fE>9dYdUEXg=E#cuI0
zk5gAdl8(7}T2wgQR~x&|7vR^CW$@Z=$2QJSkTXdarIk^<Rxud<fsGkg--pXTTOsAa
zOq`4h<dRJ~aB8f=lQlg#>7Orm!%gy4_E@xLH0#QmG45MCl#DUPhHb)Qt8`(6JU=g|
zb>@I(?yyZA!1}xD)N0Ve%cr_rVY3g<s>4~5C=S0Z<7Hpm8umf@7~$LwHP@{$(bJI{
z7p5Ykzl-z^g~9%I8>Xb)fn$g{qBBBywVm*Zo%Q&2p)Mi^RO3R478-^b^VYXrm{4d%
z>*V*6;kW{=LpHSY8OS>e?;>bRq0HvH(L`8#0m4R9Rg1xug+2J)XE_|Lm*MwXC0xlL
zz?M@w(rlg=LmHk#$+jmC{}Feg(F6FuZimo5Ls0RiCoi1705!i%`quwe+`d%{{SQ5{
zF{A~aEo+W<BhMpjpS|=UI<WUY>5vDAGkxC?ITuS-W6B@JkT4V4?i1gH?3<4F4CXZZ
zNYw67hn21xm5s&AkmQGa?+h+$0{rxTk2L=UIAkRbB{y~HdK^dUU?pDJ`4{8e#GSA?
zl0Sc$;>5T!vbVW{eqZEWyTF-G4fAk)jx#4W9mWKiou<Zh#;5oVF#lVP4=%A-F@HIN
z<gWVag|4{HZD4)37T&k*`E0f}-%ePG&aa0swL<!w2NXz>{oDoHR4#NJ0r%^&cM*?c
zp58&2JZy>wwMw}EtugnOwZr<c+Q>Unk73i}==xC`3)Q4gCLAln(_`6vvJ%ECW@1;<
z_LwvO7Xsz{?)cst&g+d~7H|gdA`~#(VvOIzEE&CE50)yK^6{bHipn>;@u;&MmmIYe
z$8sh5ZmGqBv4d#5b0btHS-{ELlrv8#A$G%7bQB(Fuf<L*+&Wr5_r7pR58%eC#%#H*
zF;rG$$iAT%gXeu$3@_8>Maf`C2mZvt8Tx28#}q3ITH@m^Wp2&%;l`*OEbr`vy;ZMJ
zy4;3`BEBmULR``4QZPD1_T&-ihaBiskH~|LIIJ$Ys(_)?`q~QyYdVW-q5yx_%))@p
z7ZITygctQ;tZ(j**c2@W8A$I*E14VTYw+VsKV)Q!bLfLB^(*ec@bx2k&n2-=x@4!u
zZNLv>9iF?Zj-1x5IsMokjJ~!7UY#zZ^nGtS6{t}?Z!zxQ)5rdh6Uc8{j@SKTncnXh
zzW4FKb~|B;e-?%L&H%WktKre%p_pkEge}rl9A-Hf8XCV8uF}~(C*7Bg#~+|wT@c4L
z`K72FEuNB39jI7l%T`@Fu~6j@E}ObCd~_qYAFqSQM9F8UHAj9`47?olkol(viyp3q
zZs2#tM&o`cz1WG3qZYv7w6)Accj9!E3ny3Z!DCNNcG)Zr>aqwd@cE+{Z1!96V?sCf
zYm^1w|77QQ{Ey<nS|2uy`l8t9CG-A}uZpzRs_?7|<m1u+>>bdQ9}J@Kd*~0v_H+Lf
zs%I*&;;CfaSH_`itrs^w383*(UygO|i<N3hNEM#O${W>)>#vWi1}zZmJ`3Oc#^BTw
z3pTE^=N7#hbd{X+$^Ld6Dd(O88vhiQ>ERqE9rwy3jWHx!O*-4faL;Ru-cA;{c)v<M
zYkgT09mX}`X6UnX1dsnCO*G?>Ww`;VwNLSI`3o$#e+MsZq?bPgJiKcfy4^B>$+s_P
zI4GY5lZ}#_I)!1D&5`M+j7y&fA;c&YO(dh0^VNc#JGSEaahhBg(1fcKr3*37o*mv@
zfY!`0Xf6B0Q(oC9m5yEZ?Ey^tm5=fRv8=c20e@!&Vsn&nKBW^>5^unyof;yN4`TSp
zBd~nF276MT;Y`OaSlT!k3Y|O@TsP&ZU(&%mD);Tr_WX7#6n?G6r5xx<&#@h8xp)L@
zmV5HUH6Iq2>r21&Gw!UOg1<+<VU6lCL>-9dyPOv|aAqGS_*(O}LL28wtWi~^j(gcw
z!orB*pxTpiZkU57%Lh=ulL0bfJ@~wM3>WAsv5R?kJiD+P)jJ(|K&K}f-*e;bb!~Y0
z$x`(4JAvno^mrphJfOkS;obQPhUb5y&Gj7^7rYKOQ!gRQU?}5v2hhvP0Y%SbFLTO?
zsdr1EE{xQ66FgBmpdD4OwdbtkwtOx;ulkvXpmMw%^A2=oSYzQ|_4eSQ;OW@)XAX27
zuV9$geH1J0z?q|Ad|q-L8*i6jWta=J?!QI0`W~!$*^EDuQ?PbrTkMg0<z(sDtRLQn
z6-$CR?O`K$YJO9Eo0*GIn;)Ynu^ob|pGo(@iIb-lVT92*D5-7{XRG*RG-LVVLvw^#
zw8FYMN_bf@OJ+}YNQi4e-7mVFZkmL8Iaky!=+85^+G6qDZq$3znXQu=;oqo{c+oHu
z=`)0%=FyZ5_xjU(r3HHEhH$`=e+u>KX87}HF1A_P@xFNr_<V4Ks>y6jjcSQ^eblkb
z)}0&IUW0y-Cdy3;grBQ{rES}y@_0P!dn`aw=c};QQ$fk_zlv#l+cWW>6<5d^W$tFl
zN9?SG#$w^UUMau<FH20Gvk=2RD&e|;GPiu2g=?2wc;mxY#mP5!(CLH;!Z$=B&vZR3
zRllQb_c3JWY2onSP)xpV2H#uJ%oz9^jn6u82aQ=bc_s$fH<g^GHHOOlqS2KkOs~tw
zrQ)VESm4EX&qKLcVMM>o`}lnM6RsBKB1v)s_2G_e;5{rW5w@esKI}Tx4ZGd}hpPvo
zRB{ndFKVz-_TDqcNM5Pb6kA@o${cJvy0vS7OWk{n3Ecv_%hg!<!G%B6opJJn6$>hT
zxWarHZZBPpp9<lT+#HJy_DLu*`-1azpJ7rHF6{4riZ9ZyUh@7B)Itl;qWdI-c(;e{
ze@=Wf-AnS?EjVwl5o5AfB2nopj@25lZ)s0%@STo92X)xmdnslYdtzhsK4hFR<Hs~z
z%xu;ZTc(;~VLv&;FYe0E@%D25SHip?eFjF@F{{~k#mfCAJb%=iTSH~;`}PzJG@Wor
zW~!S<{Y1m_W;pv)8Mpp6=f)0t*qvyGiLKQ*@a$CFG&5vL$|;#c^}^QsX-qFTA#+_r
z6ncGEjF`R)%AGgj(RvrE4DzLsZzF6o{G)i3K8%l^*)YCd4R0^BMtgcQtYi~(mq&5R
zmuJ|K9L0|*FU5^m4ZD!}Sb1A^NnZZ6{;(bgWhWVXK=`Q*kqq|ThA~x*P>*Q($+_{T
za*#0k%g|sXIqz{+oG{0NhK5d<<NZ(3yZ>;>bW~ySiIz-E(ZMriHD*dqEVxJm?TpE&
zid6aUS;JB<3t2UHFneiJ?z0=lVP${tQ^lAjKN@4E@)F#<d=tBk{rRxom>Yl9!)}=x
zntt=gGBZcKOTC9X9!9voi}Wi~Lio<#ik;1Jp_FUSRq{L7Zrm57&Fq2&DXsa^qeAvu
zl2Li^SE1t372zYTu=R9fP7qI`S9^E5hY#c6WAzFjKiL`8t$>Q`Ty2wurK{PD{aZO9
zOH~s&qm<ERdQ;?SjOC(tDttam?yK*Mkm)K+hnrKes748Yo*2L&xe+>@)5PF|;_%Ur
z7mm3pEyTlOKV}pH`?cnix-9I_9>NpOi!m)#m$f(Eq1j6p+`ijeT;l)p8;U<k&ROC0
zOR($oJe0e%WZJD%?zwsv1#evGxA!78@BfOL9c>XOPOu)kBRKPm^b=Q(!2&8{!148{
z8`YZnz1JgH-p9R~>Y%3NDPAr9juxBNA$*}D%@w^kQ96*jwx2_8#2$1>{h_$^@e2a~
zQ^GXY9(d4Iyb1H_6|a96;zg0LVba9&J9z?b)X1|=qe{H&!c#f=8Cg4u@G45dnDk<_
zc8SL5`Nmwe`WCX9`$9uSg-Y$a;j)e&T+S|p)23JWBfYUPE+_D?!#13(&xNx5oZlK;
zM%<>eIJ4>v?kwAeXNaf6uUJ0VtWVnutFWe}f}PXb(7Uuf-kB^wUpar<mi$s|)i6h7
zwy=sL6nuQyov+hex%{#ZI{wwcw3-gw=9Z7EXFGB6i(iU)nzPXUv@Lt2RN#zq0({2q
zh1mr=ta-iy<=+PK!^bW((6N@zs&q(8!tl~@1}4w=p)l~Oft|J`<8F(O?9c+~PW9#W
zufG-bH`KADs}Dm4jYDj(F8Z|lt~gfpP0`Y+J%)bOVKdtr=p{(6`(Q^@|K5%NMmOL@
z+INMDt}5HB_|Ub+gahXJbHBJojK};`WTt&nD37s{E`fr6;YBdh?89+hQxJbr9X1WF
zuzpalIOt)`iH2VlBV@kVXLf5I++K>LfCzYvnu+!fHrx{ui+xMQCq29=kF4*9ck(^z
z@;ZQ>+<&3urUqMmK7ei3L9~d?hE1U@vOlP>e|=x9`y-6yr5kX}Dhu1EIWs^hoGrF5
zMU<)^oB!0~$JVQGW$Y8Qt?10pV@%OQ_T*z~<AhP?#Sxni!J<=V%)P0K<UQZ;dE|O5
zIqgB^+*Ujo^98FtbH&NxiWl13#IM&0Eq{Jjte=?9hIIwdQ<c2t+C)rW?8*1SjmwQI
zM4{0zE*SDt(Y9@K9Ie%Y?Vv~479^hwk8~DTso|H!U&W?H!h0;Lz^heJj1P5TeQkuW
zA=`1aODGTNN?+#Dd)Qvn#NIZIu>6%mW_v-B<y6D4i!sQe@WvINFu7^~^xEYEjZULV
z*Nu+p{g~ZGk42jo;P0HRs49>g-jz9cDzg~R7m?WOo`c=DrFW}40qw_$Tjip3pyoNT
z`A4aYWSC=cRtkJIlwte+1djY1DxQsI@SV__OCgzVKgpq=HN%g%-&i$tD0PZo!nuV$
zi@(|6uA^k5Bc%tbZoo~q#<237Dn=JuaJ^#~$`dtNb4J|HhN^sjY&}XI=ySb>EoV9Y
zP!#KXpuxohS9|S3VsQ_&I4GUJQ)`i+XN0+<Oz8D$G~JJ10(F}3xkDbFKRbkjX6oE+
z@D!y3&Dd9*pi_*@@q2_kTVgt(V#{B}a*gkb-NQZjS6dmDA>sgh@dCvbcj5E5DSxbQ
z!Ad7Tv|kj>f@2PB*w>YokG)yaHJRpfZz6J{uopH}Kv8`YT7E6rSUnd_%Zu@|pC68F
z>n1t2D%9(HWAqFSZkg)9+>61mcAv(+#?yJAh1}h<>hbT%Aez@%GOBDey(G^vtTJDA
zw<UZuL$WQ8#^7pvA0~Ym$*D!7c&L3p6vnT`iF={+vW&x0=U%u!upYrvvzU8xHp=vE
zQR{XQE!~?kq{m%k_OjultlzlVay$>j%$F=mDWYq{7Z{$v@qXe1%F5=*+a3%o8NzZu
zd2Us=pk~=r+MJES)+NCxyxN5yZN{NygB`EwdGph}ub7e;&Bw$2xcEj6ChQF6)p^-`
zAxs3t=q9iYYR%DK2U4xE5j`%ebKdIl;#!DAU9}5qD~z!)6bRJb2fvk)_o|AZcYrRx
zWSh|CzKU=!)}!a040KdWp#P3asHba727LwwNw%w5&S#t{JcQVa-e?-2Nxdp#PFnjH
zN0$eo;insq?w^AYw;6ah{RW(3+fzq+2?^(?aQ0$hMi13tvSjd%thb_`cq1Am-Il!I
zeBse1u<z*MTr!XBl)nxKe%^=cj92JVy$p?20vS1`0}Yc;;kl{>H|bBKd(%t!ID7^w
z7aT{aT?J|t_n|&8mX-2ncEXuSs>qddbs{ZQd-F@2D=n5?fzyZ_%uSHo+NH7VQ8ynO
zr%d6@R}-nVIhMr^X4tc1BeeB~pmT@`m6xh=YHk9}&d7V?eRq6mpU(qci(y?mlCLg$
z@nPT%1UNdQ=gutjJ>L`)Cw~@hVmx0BHzJ-V<Ir*=j-I^<b5jPg@J<-_yO!XO=>o(K
z48ZpxLpUm2JjM-yvJ)yt%ixaKpX<UdBPGb%Iu=hSsd3Bh<}knCmRcS!#jB}_Z=SLE
zjcNRFzmOBsO6ea%=A08wuDBEyX*6TQzc~JP+W}1@3g{$vWcP}p9B1#(-HBs4W!wwA
z?VN_C0kcrE?;R2<{MhTqKsld|#9P~RbUY*;jITZ6JUgHEFPwN@xQRcG4M#UOxd$)%
zA^D8%k_!|rkI^A{#^>|Rxo%WAI^fUrLOil@#<*|ak!`9@lNK)M+NBwc^|Hh-=FZ;A
zE%?LXJ|<_$-$x~dHo~6jlbp(m08>ufn#bOX5nR1k{E}PyLFIrSPtR3j^5qQdY)j$t
z#LK%=d`JPA)Tj*Of`KvgNNXv(uu%FX>)=IvPyYL_hnyKSxM7hgZ|93sPk2EC9@=1M
zu<ZBSrBn6JLFs}TarIW|YlOPv{WSS`O-p0dnVWc%r%p9~!|m8oj-ONwk2p1&&Mjjr
zMoP!ujWP1Q^0X^t|07Ruf6+8fa46uduOBdR(MkMHdIiUz#`G&M;j14dxHq6VM=Xx!
zmf|_6>iGhm7SZVLHk3t*MNph8!qM(s;1b;dP8YM;LOTKnGCZL5PtFCa+M-VUXxW!_
zxaVyFTle{iAB~;hc+C}8&*Y<UtB2fwtx!931g+P|ZgoKl(!@Dz;-Adfit%{q@ENlv
zx?rozQ8a3jB-{i)Ua=7u&f&?dd!Wk4Y8JxLpNsUgQTT6~5%o>KV)@MLNM5}V%8jQ9
zC+;AuH)>I9(O&F|ZjMdoAL8WzaRGO1&mBM8aL}(dyrBFC-Y4{7+uH`EY3e-t`zDTV
za^`zGZ)B-ipz9_x#<VzuJhuQkM~gS|yBfQS+w#ecWG>1s<YJ|FP_Ig6zocTOeG?|<
zz&@Pg`4P$oFTv9>g?*fjIJj4(+%dzkWX3dBJk6ueLM=Ais>f#jF5<T@Vs?5c?P7~r
z*Z&sW-6k@2eMkBKCE?pqI|klypkHJLw;7M(0F_)S=l?^Soo$i#uRV>2Ch?iSBkx4!
zP`{aYC;AABD!d)kn#ns<BUG}Wx%~C_97e9Vg)y4X(CJY(YCK+y2mj7vf=6S%OzO?0
zN!#G;YAl(oG0?Z_jb}Z6;#^h->^>67fL_8SoiGmLLtnyC_SJKQ3lpw+977t_VMnd-
z$=kZa=*veux#!2^ct<EKt+@L{U(}6tWyca5wtn;sSw)uY9H@@chz4BV8zCI=hmu1$
zFa8xXY}$~AxMW9mI$FR5Z<oQ-J)iS-uEVBhLm1Y%9K8qk#9_%eZOY8TS6g9i_jIOF
zP8h1=O4)1Jb^K^4o>rT?*n596&y<K$CC3fk1I{9`do3*A`$F4x5W?2oha%_}eEr7b
z#f~@xO0I6y$nm_pWjzL@B(ssFH6Cu-jaRiD@g-Nj{<eepFs74moR%Z&t9;%^HK)(N
z`IvN4PXbYk@#8`&#*T4f$#?nt56h>@HU;-Q(vtqBJZpc+-O|OIqq{qCqt|?7u1~~%
zCvz-WtVCP)aok~a8oo!`2(wO~@0XpyF%58?W=p;~GarR_mSSE(G@8B3!UvN^l7W?;
zt?UtBhH4^Yw`7WQgK#Za9qmro!#=MucN!01AIT4xoEENs_D`6)=~2Iz5ytMwLZ(uN
zoLS1~C%NZ8(ih#_MP|bG1$4AOfvl>@be4Cwp}rv<^JQM3n8FghCaey(;nbMl=zVSt
zrcBw2i7oAgQJKePSz~c;{%`c^5DfeC*Kp;XDGPro^G~xpM(uN9lfCKC`}+j(Il?bI
znTOTJa&PVW0AE|0ak77Bw22?iW^=A#&Z|u@XKNfT?vKCbJ?OJ*Jk&qi;Os^N-dvf=
zb!&zr@L&cyU0a8Hkvc3Ho6qS}r|{^E$2j%R8EVIk&^vDuqvZMauuvCQO*-=MZ7t?j
zG+~I$mK?uJPo}RX{d~<~^7JD%Y)+)QxeMROEUxi@0;vC$oMV_iLtREvt<Z&6llrhr
zR3=B4dvkB^B3ujAVH3&BL`T_T%`@>^ytKl1$EWc5YmBvNc8vY0%Z3koFyvvG^y~U@
zv75MzF67YX`6xO{uHjfk0rg%N^Gp3?elyeHsH;}&axb5UtC}#m_h6buEX4PJA?);_
z6VrZA#fBN$Jlx6;W8){#bLnLE7nk48X1%zpDvVcl_^{7{OeStPhp}<fc)}``{$B^;
zwOU(@JMYB0#yRxt_7$Tu7Gj<3FTa&aHsQs2+#A%7_gRWpi?lh(p^vaKyNM$#kI&}n
zP%BV+Jt;2m3vi-)vOkl`wn@HnJU{x~fr0B_@zZ-^>bBRy{%On}3(}d}<}iMpv!;WV
z9bQ<w;#6WPF3Zl$N~Z$j)JhN(YKhNhq_1>H2X&FpVewcOgCmtVOuDaj=E7Hyoz$%E
zTFAUFJX?QtZrIlyuiojQZO?3M5za>R0&9NBzYe{L!%_F(DXz3%hTMc$F5LwjS=E9S
z|Fz+|zLU6YP%AdRbQ||~K0xKu6G#i6K#LiTdE%ZHGslICoA?Pbf@4{iET8k43G!Yj
zqEcQMY@Pk^X5LBsd|1Gm#7bNk8o=W9h4gjr!G`c0l${*L6<&#aWYdRdZXdu$wHRie
zZ9>24%V2dvf%IkG^v?F@zK9wOu9m)M%`2?hZVQbI2{5tF;AET5aM^VgX-DPX?a&rC
z8}{Jgw2?@f;K@l#tKr?ykL&J@m(E}?!&F0gv9+*!pHGD5?HFb^@x?Y(Ctmzx1CtrX
z9Pnm}bdsi_yEq{)w)Ep~$~j-N9MQL9ah6$dGt}eHM#`M(-y8mx1Mz6SWF0Dwi`(ux
zV(RlS_lh(8msunINF02mFBMWcjrEa6ywoXxt`k#`lKld|mW1%m{*E-wdWM%*htXwx
z3y%0Qh=1ElHuX&!Cs>ywK#?Oo(Yv@K{m+iC6Bwnj1J>G$q>H=|pFJJ;@LUWJxtxQ0
zU{`j^vqf%28e{!JP*?c^?kR7PFnl8Bj2?hC2JbL1F9cy@U*qJT&)B!G58^(h${tE~
zK?QMemCx9LO&t(+Z#fKJ^hZLkLVWvo2ai>}`29^j7sW3}@ar+OmX49vRWHev{f8F;
zxhPF)!|P`v*kEJJa^DP|T<Fb5W0qm8%?z~EUyE~}`#|lh8$18YVbkWqyG!;)=K^c$
zKC3{b|0QJf8N_2{N8qMehUb}g5g8kZ)D70y<uDrWu4wVrg%q3)bHp^2avayuz>xpV
zqPK3ts5oDQcQa*)nMAEZ265N49oUtc!=yPq*j489o7;<5>%&5L+-=9Alsq)@bHRM!
z&*p_s=4+*Ro>cRs@5a8=YStB3+ik*>FT<Ja(E-sR-aNl;1kJV<(Xn(2ZhfDEM_;G1
zB&s>9KL119uN*d)XSjd&wa7f(9qrb&;^~we=C&Kck+bqRa!)k6W{+T6Mg;AeXR}>W
zKi2&)LBF3z(KPiLx`zkKOyNA<mraq3wh49X5)m8s7Unf;u=c@WxoaOp$Amk=sF{c@
zC+iRzd=x_#*osS42booO(7)S0Ozxu#_0H0_8rGEGBD-;~oiDpw^5dJ14orymVwWXj
zdDpr(Tzd3J_M;2%QQ3&$bN3@;g*`W|8P3#mr?C8)<X0|u${oECf1GH=k##C~_F({1
zuO36`h1(dUw-!e8WH;QVGat>1V!-<j+|sfS+U+;y7#sN<T6uEHzD0N}U+2l75%lZt
zfqsKM@W0u???e4qa(f0gh7V@w>n@Cxox&2!k$j#jT(H?=&~aWTgl!K+t@AF-__YXz
z&w{z-c7GaQD8`JK06gy229qq(*#A}{OoWa0Ri3q~r^RFE?~Q`Qndl{Fjy=L-c$+eb
z)vkbNh6M~9S72U}{0wE!?`xEf5f%pg>C>2h8udqEZBs<=HkA2TD5sQn#`wr=j@UX0
z4~)UK<#Fu3ay&2U4@dl0BP{Ncz??=+@Gi53%wU^wbf_^}&GP5uAL>X;8pplUgXp#o
zsAyG*!`lo=&ygJ4WF7{F8M2R-4trhBWbnakxHu-D+}ND+t;exwXB0QK%csq>Qoin5
zz$U`@4718$h36i`mhHhB^HP?co5Zl>4d|_-%AH&5u-UzsO@d~@mrrnNPA#k+`Qmrq
z%jiDXm$&v#<R|O%cvLu$T`r0zb$>NnlG~wcvJI+aCfn`NWTs52f$5YOj$LYv>*E)|
zZ9^a0I~mA(xj&r!H(^0s7^Zej7j8l(56p^VGjlt<7tZjHh<MP+pXJ|2L3?Zj>$-G@
zrE(8++9iMHG8Nrln9E$-hqwMW+pzhCac5d1{Cf(9w{YS7%WYUH45C*5+3@aTe`?%W
zk1ul^nYhuO+rG8oh;Ejs&kN_o$}BRdDeo_-!@*bivSW&o*?BD5*BwFH{UOYkIkw7o
z>B7cp;#l!p6yEGlpXS<J+G7rGtuREQemF8Wb!T?uaH{(Zr;l|s51&-0kHJGMmJDvE
zjs~2!U6p=SCA==r(WCQ2neZnHYrlkX_;XcuFB`%G-5OK1MW!(P-of`+e>nUJqP_kr
zyz9`Dil{7BZtN^w2@Sdm|6k!FF7gC%c}B)CMOf>qxh`0`t_!DlnbS~m)jj@<!k+E#
zuyEv5cG%&^b1T%j?0s9-rfTun{c$)oFc0dLmr%Va0&cf9K~-in`x`Z3-Ilp{{$(`d
z&6{zyxIv$E@4-0ftJ;R0!>!2{u<onD-fPWi-%0YI8IN$WXacQYinHsM9S<M81*eAw
zNOsFcSXgf?o9BSB{f)ROZ8e@=3gnUL)p9OvjOI;8NcXM{^w&ADK4m=?mo37te&69#
zsE)%y!Yb;ZhDe{S@cOw65jEoAHoYN^zCe~Z<j}pk2j~7OrgEYqHgzau<;+1m|23P6
z-Cb}j-hvEB<w2jlu=r(1n@k->&F{~f|Ab!@AIXlvY4mMTM0F=szIj1}O&%pIp>r^=
zoXV!f!iX((MXz4lam)53UNpDk_phlk$B;engE*eQ+L+ZUTM%Ko1<~vO%I+oxL*Bb!
ztg9uezuO?;tPz7_OQ3sWGCt36hpy2GZcJK;?&n+bYPB$7>@|6H*=PiZxnccZYksrz
zz$?k5|2`1HVUt!u_vk^)2v?!@)j?1_s)erIq#yB%%ye;sMbI|5k5T4K*{zJe-Gp1;
z4WnU+A9OG3W5}XSFtUw8tjxCSOgiw^T}P<d^yHYO!ZF+?-Quk_9BR>q55DE#<D;g`
z48IQJ^K(%dnS^ocqnPe63q!97gL`prnk39d&UX_|R(EE6zT8ot_2k7Bbr`-XkyiF=
zu~J-`DTj(V<4!f2`J}Vc$8Pu@;!neEGP8Rg&945Ru$TGtS($|Nx#C`K7|LCmfb*d+
zJ~7{nUTu%U)%!1~nnmBC1xz1rPLHn<*eLz{eq9{c>~R~$=?=p7A>!tFl*_6WcQA6K
z5tonKkJrNfN;8jUkh1|>d>sw1#jaG-oP&HF6*dd+%UR-3TIJ)**0W`Abxy&|pR3Vj
zLK@Ep8zaqmA4VM3W%>4|xU^9G;VZ;FQmF&Yu>Nue2$Y#b5n5A&fq`+XrEo*l`eX6h
zk+d_;M^cahof5P$J29BQi&VsE*$3UfcSO%xTRtAC3e#O0EPuQV)8(BW{@)m03jGfy
zMi*dHQq2BJK@9PV#<%18EH@dz<#Kn;=+c1=X3aUsqmYkV)nN8s>7uyL!9ICUd+n9`
zVxl!YR{C<?<1uJyl*Id|NAaBMR3z2pvtVBu`+DT@>472aeW)u3X^-T>yh+&Uw-{BY
z|6%-YKU`WdiZ^40X|;J7dLB+?;_|kbw=@U+$3%0o(P$o$`BRnfM^}!Uf^OOU5ZPis
z{`!q!LF=hV%e4`Acs@!cXJd6}G+xK4Qe&GxRZ^Sqv3@QOoR3Gb&jnmn5zp$7jrjR)
z1V3m6ajj%Wo)*T@Meb$ARZZC{A%;UbYoNu&cs8jG#LIaxNcbvQ+k+!;Rd(Lb`nP3B
zT}#ww<RQLnD}qz(s4N|?RsB!k`?~Yck&MOl9&>S^Tz<bj@_Ajl5l6a>Mo3{YjJJ%X
zde;;b<OQSKx!KtJuMroRRAPKvxo3N7Fm|3~9{l2{T9d?=idnd!+J$eg>e2e07iP4o
zMciF|X6{L3K>BMeTy!1=TG9AeJrv7dw&!%c0?rSdiP9k*cthAJ&o4Ql%x)4@77gRX
ztlKD)KCZ!<DoorWcOkcvIOglcO9`#SVLcFikI2mK;V&qiGL=ldJi~GabM-H6j$7o+
z$Aux{`WVWC>X~@B(~B!Sb>$wSgEvK;P;J#6&!ctZOp`~I110$MuL<_;=t|>ts;pf)
zh*fQ+ulCfC&6XzNS-WoHll+65ns>10rv|R@D2A<BiDyU0VB*1ZsFA<ltd>K#u7#^`
z1Xac7atoKmXZdTs8S>tAM2CmceN^~SYs@SBcC5p3S=Zg&JPR53bNM((GAtX+xL3T}
zwlB3gE4Vvmb}uIGY{$Ye`RsJb17RKQ=z2Sw+wY3Ic1Iw3v}(uvvsP^My{CNMbKs&h
zil=>=@>H7>Fs+`AJ=<==@S`6rLVM%lZYPeLJB%0F7O-n|KPWMYarf-d{!Bbw3#IE-
zART5O1&@zU;~%%y(C!{g$I&D3_;P^sDW7Bc{ZRbO-;ap0T+HzrOeoCJR(=j+-x~AY
zM^k9c4W~-&Xs)~c7&nrKapO4~-YQJuYg=I$rG~P3pgsS{=PObePCBgy!1!nYwL%ud
zrO`k>w9>?l&K4-Dmn>n8Fg%8(qTV?H`jr~EuU(E=iBX)c)0%Y_m+-S8orXWxqRo!B
z;@oKh^?DO7GnKpNsNsCNRf*+mDiL*l1kW~DP|5uN{~4D3X-(b0Ae8*_!2UKpk$zwr
zf8WmI^2*K}eQh&RB_r^&jbxvT0`U4*A@f({()eaB&rOU&#^X`6^N_sSz$<8&SB>SP
zx(FZbKIS(`XOgWq;uh>fs&Y&C92yJd{Aoz&H=fP!_hf*o<WjG6rOS;;eEwzvZ@oN(
zoV09A@9u~D&n=KwC!U$@;))a}k7AQHo9L~<)962__|K7R^ULM;WElp<jpn`L1F%j^
z6h@vB{Tn(n{BsA6+VKGOYiHopV0~OZJq!saC1<1SM4Kln>~?23TfIwSu}d%Xu8{0-
zLLRdJ8PWc~HP8)c&vuLE!}U!BK3&#?l6@IUV`{Lj_Yh{)v|^>ql8kec`SJBiG@9C&
zHG4y7GFk5516y;@<8pMK(U*5`_+d?0JndRl!9LoE-3{yT+V~tMZ1v@s6_?>Rd@5(h
zyK<>b7>k_?Ick6>m&V&;OKb#R<_yAW$8LxlHJNXRXRzp27Qgvt3%g_xqn*8JYFQw&
z=sY_6s`IhcXnr(xfZ322h`zJ|4-dw%!?tG#{cgoSXKm?eIE064Z{d9XF4SxqgcmIu
zkiNoN@_Q4+8|co*ZTfOy$L8!kem2YwIV13j4sTu(*JYM4qoTVr>ANS5eu>9TW^B*w
zHCS`K1(&uQNVgfGoNi;nyc!RDxh$TqRk5gYvw)hy53QDjGhTXU!&O?sdzuCg9~Y);
zFF$c;$MJwl6YgK!h<2V`@%8~}=hlVR)lpoyd_IED6l3-=T@<Y><_&K*Hh&+-$DcM}
z%J!-3Ghr$Zht;5O{aAYal{3Q8_Do)#$H?gU@ZI?sZ!%i4;z%}Ep7uob3}GJ>rpeB2
z9AEAhr;V})nr%&mZ!cwf|HxyZ;Q*9e4q$X?I$PC*@!g%NOp^0bO-MB2Ohz+BO_OTB
zvS8sTtKMk^v^*Wp4|syT=Hiu=o%5za`Ruwq28)YcAjv+LL3+_pvP*|YyzD6h&3Wgn
z3+|4cj3m`DIJU`_=N{=}a*?TIQsmC<w-M%34ng^Z8vm;}jVD_aTqh0>teC|1UImzV
zMf_O@G;ub%3VnBk@M>5XLx*<9OQqLnk+056QJwK|OlQ7p_71C$xX@--GralxA53dL
z!+gXb9+a;6R_(JemvhiQcYn-UXo<xwrH^I2LYNaKoF4ug^YW(hbIWP8m^O$#Wnbge
zQilO*)9BrNinu!tBjR^H-Y-dJz{RQDb-6blcY2317O=KeDXM38hpMF~HEZ?Q<orZP
zIFeb7g^{o+n(y0o;Qw}ZF2d@{4xNrUrwWj1Exd^iGjZQKjR&*$A~eH<W+mb*{_+7!
zT4m7t>^oT22C{m^Hu$d^!5YrP#(vjf6WWbif`#Q=3!KyK&A+FH<MZ}jOyBDS#hN~>
z3{&G$$0Ml!Y|LKoNAmNX>(YH}!4riRat8l^jR9T7K@x>G$~Ukjau*t9DslZGIk)cX
z#d&Iu>^z_ahkE&=!x&dQtm%l)E0Vc=aU^HNM<c_u2?{0#F>8=D??qW6#`z!iXeLrc
zSg*E?G{upz4}rJ$Vrz${Y;NHUg~l|N&CZ8ga4FshSN8QNEuL9)3=gg3uFk16O4H-5
z4TW4Muc<9>^5|~kgX62F(pPqU7MpaqZpc(dymaKhA9=WXtqqTUID(~H+oNmgD7O13
za~6*Pj7>QT-^nMiqxu)z{fHB9Zi&OS6%CKc{AN`S^~xtm@6n!W?_^)8I|lgJgsw@w
z*tzl@Mm!8bZJIt#H_zgx_a=0`8-qW`{n^g_9Paja<z8=VG+*0;V``+UzHS&h4I;f3
zhhaf!bBwE+z(s2JFxW2+;W8_0^{*Gcx60!1^Nl&vOq@P*n`4i;DK@F}=e@W-e6uB;
z4i@tDmo3Cl=}FvMScCOFCZO`t4?OwWoww=}>8jcngWpOf{rpC3-yvB^nQ45qRD)Jh
zcMO=-AMGv&a^lV&3~gP=a-|grIiJqqCgKUR(q^`5Ehf(&LiMm`a80yewc8ZB&m4-{
z&uw9Sa~j8unt@=KG@8ha-lO9>+>+06eqlb}&K$~C|6;`NExzB|H_%mGoPkfwp)zX$
z7GJHvzd3{DE*!w#Cp*Ej&H^?wN>L-OsJN*zFK^XTX0TIPn-tEp0r9Z;Sc)aXh~fK&
zbJ2q$Oz3Dv?Nf42Tq*v)=<Ue$GiG0-U{1fY21>s}V3F)Wr}{2vv@?{BA9?}O&vJ2T
z0>7sEn0h~+WfMlRL)m}W>}<_7JCf)&ErPZ!^I$(WndiiRFx;ghRaN`JZuKX$*b;@V
z8`V(!s4dSY&cUv%Q3%jDh$i=<I5y`3j_+&3_7<8rQQruw7110Z^H{z9MeL;Q%XgAh
zQvWEkc&8jjc%DVVz7$5TtVZPf1!y%%gW6%@7D*V!N!Q<?xy%#|*9=6nkxDFWHG--|
zzSut|l;yAMku6@2l@7u@bsE5esE#P`55X4Kx40B+0dsX<X!(?3&bL6u_DmB7Z3ep5
z_+WK&dtP~-OKX|kZN5DZN6%-l<HsO2ePYfzU%IgGwL5sIG#T^7zddV34?dS!#zJLv
z-tUozUg@3ixxE$V8l-b$b`sWKdjcPKC%)ghMtC-*xaJs0gNG%kwNIwS!Li(a$CCvg
z#Z~633F}nykT5OheDBXDjpaVjGoC}MO0dq@Njli_d)QEhTJgx;=)M`>3~hP8P?>gX
zEYL|={F4V-uxz6WwpzS{^?@$Da<>(|B_DIS&>OuEwh<OjA)^bF8E<?A=RXhUk?=8$
zD4U1lNzT0Y^#t}@lNqyrFX+v*l;`Ov79=|1$DDZ1P&$gDFYPhbU^J`-O2>NlDE{ts
z50?INrrSP{J+4P_>mX-Lh-@$3l*x>|Hw}IE=3{5~B-YHeMV00ktiLxHHSKb_V5J{w
zn!3nd@G!!JU;K0ZNFH*GhgVD^)^3@C-<oZNSuzdhE#@L5X*`%>%2zkm;G*$woUznH
z#wc+`xeQ{7XFjHXY{%_+n{eRGD(KZN!cm_Bya-P~agI2c;yq}S?uguLIgB}*!!@1S
zqS)F1rz=94``DW02l8=ErwyXFeZfS_w(Jw7fw^<MkoVe?1B@SILC8G(TG)jz1Fd-Q
z>jX4u>C5C}{duR6_?~B9!`7!mnXvyk;@f%QV}5t>+YaOX-3fA*wBxpgN1-(*5~@pF
z@y*Bu@qu9+c4jb|mycqG^k+tVdV^e*V*Gw0IUk#DlKBniq%B35yk3bz^D41B&78eV
zbGcYDNuJG}5fZQs*Q7&|?kJwf*9Vc+qAO4ASdN(2*N|6n2<cS;cvd`^%O3=@al0Sb
z6WJWrj|}0HWWygRK%7GcHnjFb!}~ax94bYoa5VP(^<kM&GEa5YV_52R>1_|@gXhAq
zyFL=llmEvMGKA&9O}ITrm`J<rK)GKgJQ~Ell69MB*b>o=TQKg2@TE2w;H81MhA#|3
z`u9z8zIT@n@kD0W32UcR?yD<0!hHTuEDuPb)1T{@y`&!J9~v>UxHS*x+p;ic0~(gK
zWI@PCzE?41hnqc_`(Yrf?+_c(I`N~-Ic%nlWuDU)SjP6|-oyHIkBEd}kqsBQX3)_q
zQZh?HSS<{?r`IB2J=9v9C{xkb<vwCElQ<yjI366D4u_W6a8vf>!<4P42yTOM<7E%~
zGMZD<9Jy{|8eUtUL>-hlBEpd?)xY5Ct~ZDq-3xYx!||^^lD8_lVAWxNetVF>h%JNJ
z)>pi+MgydW8IB_n@_lG0JfvZHSUBDY^EDo!L?sc5>7o2_CthY{jj3yU0yT1;9&B%l
z1OLLsy?zdQ!L#wnF_Mcq^cTOibXLbdgOA=ToG!COM3*4u2TJGGXcwj_8L~y_30zbJ
z@zkeyywqq6kFYs-<+A|QCTdXmI~vPVgE;0}Ypmb01^eH3<bdr_uuK+)uAK%BeJx;^
zA_Plae@gFX7|%R4;K@rBI1m~JhjXq}E|y$GN(|Rc9EnMj3u!akoB{LGv2=to=M9QR
z<=r0ab*Toe+7IGmcVTX=?#JmP8?kK*AEL^fXZv)-vOfvDv9mGPBt>w6fdW(H{r};t
zIUdbQhrd?_?_?EXdO#V{_S&IiRspU4lexHo1I<%a85*xn&ld_D;yOImNT#jcckws7
zQg@3d>U-p1-ayHnKXO3V=(i|N=s-`Mm547iqN{l(ttIPiZ`+&|`+kTIXgo`e|G;yS
zGW)zAi)Npv;m6oCET7vQCj#2>*miw<yRZ#8dI4BH;V>>PuZNr02l)9v$AICdP;=Fb
z%i0~oeaZZ<+xb@JTvzesb4S^`mf+geM$$nmLD^=>W%vA!Yn_hZj$I&{i2KAm=^ymo
zPC%TV6Rl<^v!a<W8hd85-q)2qM#uBfux6a$BwQ<%ScbVwWVBBV?QB}J{Lp1g-+c}1
z9jxh9*n!^{O=Sx`@!ZsSiSMg5j;1YwZ;#o~tk&ZFUc<Tcb{F<ZtH9qRH>7rmWvTq0
zg@0<yo1@L)aM+afDmT%nYzA69>BjN|WBgh-O#0GwFp9j7*IiSYvw8<6ACx?vSt&M!
z<g-WU0)$zr^Z1I6T>mkVi;Kqby}c*8?)5_X&W<SAYKvP<a*(s!j*iQm@%n-W8cptr
zpBzhv`)#@WXDe=<94CI@a6a>tPJnh9elI#JEV0$F*Ok1QwkZrs`^ed+8G6NbW>m^d
zM0_>JmGymS@aPmYzPWO+W+x0S_k!6V6?pq(K)HJxs@jDiXpRr>znYAmUyW%wA{Lv1
z#E)%00=L$8#{6ZOIP}Q_l{uT>>0yVw!yBO6Y(J`2OptR~AwO0=L9jflKW>}E%j@Hq
zlsOY-zS;QHH-W1IV|e6Z7ldf}v+>9rPPu#le{?ONWROqQX=eP-xE+RQNG5ZhD=UWe
z5g&RcE51hYmrWM`sl~8hop4Lvc4LPUVQUt)5yoOaE;`~!?X$fZv_>-aVcGoQx&Skm
zhOke`3e5E%!u0QfsBPC88k({fII>o}XIl7R*aZUzYC`9F0aiSIgsE9&*c+b3qJO(Y
z<kc2yE1IIgTocbijU;zagr_Zauy(mRMvof6?52a*Z{7qB)3^b%$Cli4)}8)KPvVf}
zH<)PK!#(RR)IE(cR=ziNmG)SzRf)MTTO(^-06r`m#bXoKL3`IgbdRV<P;Vdp)mV(>
zt-_dTTL&fYv1~K(3MN!L((&y?hW<+BAInqlj+=w-vImJT`h>GT<$Kzv6ao9L<6DbZ
z#1Cx1grAwPuFJsDL}8u0zKL@WJ!t8e%B$y$*rwfflo?LssE}M9idqi0=pNYAsxPjt
z(?i42R2=J<g0~*t=%ZVMeY#hXeA*fwvKw&Dcqv`zPz3)pgy~CrsIP5>h7;uo|D;E^
z52onrstuJ@$rx-l1d;1LV8?ZBT(VK6d95j~`M7gfZ97h?x1)WxEWVU8X7W{gOzHX<
z4;y>)!J~^9<Y3O5uLj6`{uL4eb$H!Wo;4d52qVv)dW-8&x_KNYUl|HFvqtonJHPQB
zcSN-D$JfKXxO-~{Y#bU4@4+Ma?b{=Gm2}4U!fd{KHi`P@@_1(Sc#N@`iqKl|I~-7A
ze>Y<W2j<eWLj)}@WHTQ7Fxn%9L5*Ib%q;-(bK7(D@K*dXDF|wN%5d*j6u(jK)9P)}
zIKM!$V}oe_djd9AkTWXZ;A%=LpKWV~y-jr4rD_r*?UHz;vI{-;*B~NgJUvp@;Kq4x
z*~Rqch!$=r&In@dE$MYzcIE-6(ZWu#qDqfPm^n;^OWwAJ;fG~#b@79G%oJ)|?2f)!
zpP;*UChE`Yz;}lg&uxokl9I60%7d8gwH?m;rsLN0uJ~cN4SN^oK<mmatVt23e+LJQ
z9B>{>eBPnBJb_-PUgBD|c>bV+@dI_)E#8fb#{^L!S&fl-t~|Lp0H*@AIr~3TdR|Fl
z+vphjIdnnSZ2{=;y*J-4?Tax;=W>hId@A{y$!TE-nNQ5RnSgl%5^y`FGoShw@S>M+
z`{tOSrMc{r3W{(s(3=Kk;rurJ4^qR@sIfT*d-V+YUFIxv%SSL~k`o)o^`^i6XpRla
zMw89aXqAyG^Kk>rv(w|lANtI5l>7YhMYwqDKXlYJhrta;RQ4UnqNa0UF*^_4%4LUF
zKbT?S204Fs4{p4(LDL^?SyI%FAsh7Zr^1zwJMO?&op0!VX(ik*R|v~@HFAbJv%`E9
zR!KMOmrXDS*ac&5^UwI}<AH-yvXH)J7;+!pMmNXrXi&b7g<Y@VLwX4NwKj)cHn4uS
zDpsFWMgINe(1>vWwddp6993u=RbtVlCDQeG;R`o&eqAQM=xKYgMjX#4!o?Rlx(!UP
zBtz*+18i32(riZ_$Das8z@Jha+7`p%!s}=>-WFrsJHv2(IQGfCSzYqpA%`vTZtfGb
z8C(a`9^*NweHs_!xcvWIzG?f;oU_`F8x34x)U*QAq(eJp!3KEU3E<!z3gN2>kD^hc
zoV9e=ud6Eejado9q!Vc9+mVH-){Ou13FjrBle(ZYUEd#vd5|53jL2itn`umRtj6Wp
z-uy9ih42*n@QZp6j<*~vnXV|h{~my47ym={-yMMUX>r`Fg>~nX*y7iWv4gt8Qs(Zr
zt7oA1#bGqN(o8tB!nV3CShhM3TwK)ybM;$u!!t+J$lTYaZ#eZ#EV1vL1x;gg@phQE
zFjUj2ohhu%vTkr_x)SF+7h{!uBWABTh$|n`nX*qIXW_1#8FmHA>UQ)=@?>Rk0nY~H
z<AQLYymFE_tN#QZdn#Ghr57>$k2mcv=1?cf0`5a5aH~=Wdfl<&Ps2ahQq~+%V~OQm
z{-Wy~+08mCc%{X5{4PjG!&WyMom+|l=N3UrNsZ%uJmeiE>`J-2+wWO~`>~yIVC_<@
zYuk_Ah50*3(S=rL-LYr)YyA7rl|k-_v^F2dGXD(mWhXIICjvLjMss;qD`e=({(G}J
zw})uqNrLdChf3bMsc<!$%2{NtA2-esKiYwC9?JiQSK|2?&^n%J;>~b5VFACGFqqCS
z#xXfRzg+ebJ#Rn26(wu#?`_K)F|*L(ha)${s9@0WPZ+v)2g<AbFw{#E-XHY&*i7cD
z_NHw7P8n6wpW1ifG=5F!L@!<8!rhWtZmTA|{BIo^hA80gau72M_T%B6SLi5TcXi2M
ztQ|R=Zd$@UvTVj?g{j>3vI8sAJ9E*iak9S)!Leyk{P*G<+MVwx9X{zUJe-aecFPd!
z>_AOJ;h;=;1D(=D-aT%{!b$Csyzm!Z+%ty8@q4)M{(lsmcU+JA8^+svr=*>xmL{pL
zix!m@Qqdr#P^gR$QHo?|uk5|a-h1!8cUH(a_W0euf6sXxM}5D~^L(EBzOMJX8TiNo
zu`8@F!r~G}D(*q)3S<23{R^u~@+5jc3l(*1kr?5Oz5XAt>6jrBN*6;vUx%Y^%kygc
zIF#S@hIOtUGsd67Rhc#ZThNnAKGGR2vq6)KW=K=MiG{S~(9I<{m?GIVt3bTjYJfTU
zt+Dh#Jb&xCLG4XXo*u7_iSr(yUAZ=wRt#p+0Cg;r9jEL1^?1H8Q)cBd5AUi-Eh`(`
zwKb&Cmm3&#vsyd}D^cCK4VUW}aGTt(q?^kA<%4)pIfjdc7xiXDG8MIbXzCI!-qs=}
zc-5ic^LXf$4xyoZE<O9s!0vtFu<BvX1#a(=ulfWr?$SkH8O^cFy2>@w6c*P21Np4C
zUG0R6*3q25s(@jID^XP30t(|~A3wth_s18>-(QJuEc57b{JXedJ7d0aE<J=<9U~5y
z_g57eF{~rLtmw_(ZJVL<t}#r%ZH^v$vf$pB$!|e7abcbv^@79sR!yBxw3X>+sf9s%
zjw3Ll0q-WYp{~qNE?)YJn50b@+ou&ZPuNlA{2JtkUPXJSFitUaq{hA_7`nC-D-M`&
ze)J8LxI0p5nilfHvW3;EOx3%>hmbtOz308z{GKb<#$LtA4`!%VJ&H$tefXuFlQ`<_
zFi@D-)p7oe3wVjlU9I>?xN^a{#Y}l4-jC~-B$xLJ#zuAcz0`}wrVH>Pt&sn`^x0^z
z1_3S_+;YYf1+JekE>7P0ZahTiXS-o*T7|KB4S4f8kSp!3;`_|s*n1|B@U4)Zc{}v#
zC@jWkWA6MC#Qi1(^r>x)L-sP~sN95k^1E(j_ZoXY9KyRkvk<D?9KWL*(E)StD<+*9
z&qvZ>k{`dXYKNRPO&JnoNS}VzSf!Ip<%4-_SCPl!E^{zwnHG~Cbj9D+4w$vG9LcYe
zX|hx1@0WE^eYXzVgyD0~s4dD{x*^569nE4g`0a-p5`$AQso6fb9?E3HpH;~CJswUK
z?P>VkiEGL{#K*3McbzSGwqZQJEs8=-un7*wZN<EnPq97mAUI<(>K-IxnAazi4-wb!
z9|bP^l7a7SWhVctDd%k+#&+HbJbbP>nj8~yqi~+{n+8H5su8uu3H<aZ0CP`lLrTjr
zc$ibhc-?aDO6tz<yB&Gk#Q|TejCt*6Tk)p&v)9U%csfY-xld)+^)7(na(@2T%7;@2
zWr_dMhq{-iW3aCVyXuT$P)BF!?)h`#_YWwpwdKT(p6EMzAV=(0;>Z2q9_f*6J!#E?
zfJ~;hQl@2;6<7Fo<`~OEnC8$Q+wMCc+NwXbK9#WJCJ)Ya6wcEA+b|4Ugnc2BV?XD?
z2W1wFtL=gHbA7n(s&umZwc(0xrf4jVf!2R-VC1dL)1HBp9~E&8b5N;i&5^lYu-n^(
z`lc@Y6kyLyAC`(+&JY=jp3togXXX9pnBkR-dviPSsN}D2#GFNt@({YINIxuZBTktd
z#0oVx-03-tkJE;-V0<w2rpWI7wGs2hgL9_64SGy(heD$Y4AMJ;%FHIR?<?W>%s+6y
zs>(JCz_*eg>~Gx+y)1p0XlI4xI>T`PrY-v%cB6VTJFdE|PTy=rJ|55uo1o88l8KD&
z<b@SsW4WO)0lRM};NE68+RBV|Kwt>!$C_hC#t5GOVnKtcTQKZ+5!)#0;Y3(Am$WuS
zMR+4xuIR=&i@UKC#v>*=fMyBG9RJmdZEt@=QU7yz?Ha+ZomwE>XE1LySo2{)ZyG-{
zL@S?SDCPP=;XnokXe3}!m^JQ|{KBJZPp;_le+G&;WNH^-=(0p!f3AsI>35mVszKFJ
zeKhtEPVrR>c5WVm_hUxDLo!NRG6HE(_yjX%2=C%tJWsxC$rs8t!Y5mfbYb-VlU_-K
zFn-NG`?H2^#Qko7p`L>{cAW<+?)kznDuhAD71=D{w=jO4pk-l&^iNIjzI9WYZ=Zsa
z-<C{Pm3zf(@ojtMapkrGmPhrVR-qHBs=csE&k#N_me~I;g=01u<8hz{d*>|1g-hbf
z+A5xe`rYt9>CNn?wfJwAA2<2v$o)VIzdkJivr_2n>4g7=_rSg*CU7~p96Gy_kcU|D
zUI_cI?G$vEtp4M4@RL~#cbQDW^VtV5qqHkyADTkRF#~ZOBl)BK6f81NW^!R1OrM-U
zz$6zooBkXNO2#AbehN}T&S2?|TFi+(j;xXGG5B}~d<}Er_V^~e`RP3rcE$3}{Ry}y
z?<=po17z14#0$qy;$h@aJ}#5{xw|bF>pI|d>vwRCv%-^J6+GQ0ncCtfQ+4u$R_!Q$
zYm9~UTRR+ajb#5;CG0u14_6v}Lc1;@nBKn$FFzPf$Gc~tG4lk@_s*h`?>rb!>PbDv
zGE5rt1ocnNF<WT>pKv?EP170jA&==Ed*Jf7B~~0Z!Hf#YxE9|>ab8E7KJCn=3pyc7
zC6eZvS@hc*fz|Um;>?z-s9M=u_%h9**XtI(+eUI}R0lq35I2creOGO3z_NefGDmwB
z9S(%+js%X~r-O+mpV3PzfLpQ+adDy*e?`4U?U^(1>t2qM`$kx@OH1-(O?Yx*F1_?;
zpsT&O1h*zIbY(Xtlsv~N;g%)#3jv!&$@gE4e<uFM>#sv_V*h<u?evA+!3=)gvmXo9
z?&9-n`5t!8W1m?O;#jod*lRWTxGzJP2ewH0W+h$4bFjVo|1AAJJk0!tsb33el<b7b
z(s4MLEVG+cR@lC=6=K>;#-^ki|27rS;aDowqfg@enVyW@`5d1N-@wBwm{;F6qGy@`
zT5}%$Xi0wQ*ez&AxUyxip*-VHq1hir*8SRt+DJd%k=e7QMsIrcv*dxN=F;8nfHNHm
znfT&5dOPQ0NsF1tizDlzds9XBnun;%_v<`4>R@j^onS=einZ9USBaMC>rpo|p87@&
z=pU+sU&Z3k3bVn;|1&cyOwi_c43F=P!=&q-v2dswgS)kawtoSK^iaj5Jps7kpMYLR
z+Obi*hBrQDaqmSR3|+4Uop1fPxwZ;5wQZr{VSt>kuTd@hr?GdRVzkd|G<jD}7pJ>O
zIFN{hN0xAJ+6797?qQx|7=vC^u<2rTelLHCiM8(dT2jat_dC*YQyOP)?#~&L8Jswv
zH{GQh*tkd-^2lY;#6XVOwi_dr%Q^7#J(*zxPKP7tVPL^`XGSx7Kmvcxy(#mTKxV!(
z;lrJAoY&h11NQrKujy<YeDN2ieLJK3aT9Eou1UGDSVsnp!_lyEs#@H|vk@vh^4J#l
z5;Wl5TA8mOhG6TL4j2{KjDtt@gQr=4RvIbO`EGBym;13(h+J3V8n~IM2ERS(-1J3q
ze8LFvU!+3S8MmQ5swZ8VbmZ@qbHv%Y8AVw(SZ1HZ^Bsjne54xB?B1eZX={eudV%~O
z7Q&4j4$o>0%-G)-U+S7;!09ZQ{S9NT%m$X}c0*P>HHLkt;FX~^IR2l4IKGGR=8a;u
z*mDL~+cw4Ana0u!AIv3T6Y)Y%7;BdY<K{D&Gw<ob`h)<ipOL_&3m?f`&I)^rBwzOY
z8+xs4%DUw1=n~<`Ph$tcv3e{=mkLjGt0S8XFJpa-0_TrY=8QRdl8Z0lkpl@#e&<Mc
z>9So2--0!g;s0(}kDH|*kzFKnpm!@_X3`c0>TZ|>ZwMd=YZeH<SA1~~5)#<TGMziM
zG@$+SDJ~})@^B|#$^M-bR+1Vrf}OD2<vtdT@#e<ABXPTlD&pViQgdB@o}KK&gF|E=
z@Yk3fq<>?QmCm=_HskYlb$q=edxfO`utjo_-)4&2S0|XQe~DY}-B=uX;m1n{n!&lq
zgL-R&IeV`-Vw2@L^k+D){A`NzUBcjCq{az~RnWQ@kK47~;NBt;(G??@<5b3;@}Agm
z$bv|Ji{mdW@OOtP=I;52{biQiviJaOg}*m1Dv_y&mO{zh6rEcc^KLs;UW)F<*%w+P
zZ*&@~^CW}&q?{)r*5Hg^JMO;Nin=EqFzS6d+Pd{+wnJ0=6?gA^nP-Gnwv;@12%onO
zW{6`XpSE_!^G$BFiws6`jpRn_HBgtiNxE-ssOGv7#lmMuuO3E979>o1lFi;YvU<2X
zi`^EXew+s<_Hv<iz)?AK263S9JpWs+%wUBTXmWTQG!AXV<VSaKX!%v~gm;$bU`OcP
zm<j)J+tA{>Dl}f+gTvT$xLsn8`<@o8ztIyj4E1^3FOETnHsMdv6AWl#gM(hNP#Pee
zG_4?9%<IeR|I*QU`6t-T8^X&?LoshgOYExOhBXhmqro+v<Np-#o<{-qr++|Hd>B*g
z>rnMr&Spj`l4;wEuRB6uGvzJn74-S>c1L=)zJ@<XteDts9O}!)a+^*#3odtr#uqa<
zbv=kHv!5fcT@_+9Cn8Z8Ig#&6xhquM-Z7FtXM5JTw`cDGW6?n})$M}hTz1odM`kAA
zXX#`-i_U_=dvBPyN~gW$eT*HQKpRti22BWN=gGIw>9s14sSM_{5^>5t`-N9?=g2(e
zG;%fKxTD$-Gh@AZaqwCM+!EL2&QSboy$$QGcBQvd1RE=54|VxDKA&{tP4~W>xh)yR
zGaaBZVLC?mw&aQtraV#M$Ni>iuxV7o-=<}FH+~tGk5`1Z^v&)lZ^K|ez;Ah1uGWa<
zppB(WNiczFN9iX0^@5KuP5f&05ZUDg4h(W(kEJfCdE6DBy4i6+j1htz+n{Me5q!>M
zvflCvmaOg0x;riK!Zefb{YtTTUoVVX63+7z6H)UlmRpvc5hnIPz8~=rbvvW6&%}`n
z!s?(b{g%_=;viZ)lEY7_almy~9=kan*TXj9m2(K!C~6_O$7t>>u)vfZ!)T$RL`{7=
z@oO5QbeSGyIZfBY;p~5^0}swJgOYL&q?nJzP{}Mu8_a-a$3U!^HXFBRCc$Bw7KWzA
zP+vzC!?TWyKVFq1Tx>CSa1cC-4XE~VA5L#Ogaf%Y*t*6R1=+sTSQv}KBSUcMUMM#V
z?aDT<=AceyI>yEBSZ;d)zCJ1_?-Kx*XX;Q<Xh)UVmoTtqFPxvrjJL=FPg7su>)g@U
z&{RCjp+h<2RUpPTv4Cyjf0+Nd3Ct|a@%3#qOP7?fAzOL^7l)zX(GNViyFuJr#Z;L(
zm?lpaq2Rt7pSLYw%SI*6R+=c+hZ5s1CUV+WWBl>WWa4wl{N>c3+FG3V6>0KLn!(7H
zsW45{!H6AQaD2@$)LM!M_H`p3cF)6$5sLh(?~c#`S{yxXIa0fY@cw@-(R8^xC*R(H
zwZ*{c!Vq>CRm^Kf23)XA_D7S(!bP4lmTuPQ2VvR{YKELC(^1zV3iDrSL*<Rk#l}n4
zCCrVJek<dsg(?Q!d4dp2P2QXyh=vIX_!Bq+`=W=U=Kxzc%Dkyr>rSlqZYr}-N6g)&
zh+F>oeBJ#Tj_(_Yv?u1+|J?-J4h3^yeRoc(?1Sx+fmpLeI5bhp{LwF;3iE6**?&HY
zWxo(}Q<+L$J@9%_Buvj)pyE$FA80tD)xz!=rCEW~PkQ40urkDEC9_86em1h(+j&->
zXF88YbAw1!P2UD<xN*xCBOEvVh!-6-sNY}uZYQ*$yZ;SbXN*Vj{X~4bdl|9fm3s2K
z1@l9Uv9Qh<t@<USA-OAkBAcM&{ee84r6`$kYkJxQako+(K9AO6Mom7huWiGOn>ASW
z*M}w2r=-VFRCN1-m>oJy)3E21l$CfW_lE9s<r#iBo?a7VuCyiv%C5m&^WBI0?%H9q
zSr$gc4MVByN0!BhW7rEN%sP_Itm~a6d(synzFj!*hb63=JVp6YU4#Y9Mb-Qstbd}6
zp6N62@O5k4h(3WmhdQ&py?n>B_hC2J!%?!oDqGd?yGVgUlAAGci#a@xenOx89PE(p
zPa9>)I8`)3`<~*0?W)6Tj-9YhSQxW^TCz{o7K|&CF2j{DInS5y+l7PhayW=>!w#U?
zH}NcsL#?)V9}Io<4b!&HLBZmSC>$Vch=b3ty}^|#wJEspIe>%myYtg@KVGqF$rhRi
zaIz{ANfpJC>oSM-4|6cKKWi`n`A;G+{Amy>E4uP$Xbjq(FoLsX5+BX?V0yX>gRB?f
zw{vg&uu+3vsx}V1zlj}oDfI5qOq`Jxc-~T-C+2*>Ft_<=aXNs{(q>`n6>}axBOIiH
zyV$#-C%4WRi5#nKu%DsJm96}l5uL(YA%U3Vd=d*I+TmfjIMOfruxgc-@SpwZd%Fpa
zyV}UkDUAb1Ct<g+XST}yZlv%#J`L^0f+f+iM-T>LHzPa>)Z+u2L3Gu*g(**(vrnBi
z4yWg`vd)#E*3r1RV<cxi9LiU-FW^R>1nwIh#eB<)m>^Dw>u$oK`qz=^o6WI)Pe(2+
z(L%Jl@Nm}|!EaMfeA-}$CO3y8{!lzmg?)j|DPQKU*JAtqGok*_3IX8((lxq=&3W?O
zy(>P3=9(~W^&2CT+M=(1Csx0T#L@tDSk&5~!~0B{PIBd>Vn>!=Oh=pl>fw1s`c=NI
zF)M#9+#CGy-KH)3-McPaZ&mD?-wwrNB6#q3CL5Q9;x*@DtuRy@%tj$k@=@<D#7H)5
zI;LJJLFxYE=rLIpe^%%5*smx?M-If`jH#$QV2Oa8k#Ny&4|BI--do)Q!Sl<|=TUzy
z-E;&_9n9c9Vim>)tw-UVcAR;nCA`xALSJX0oWHGM-$Hm69@2^YE=>RHsp2&VV%G5S
zsLD-YK)(^3YH|s^A9kj5q800WYVm5_OC&Ueu<Nd7-0&!nt;FT>wn%mfDcbm$qJ{k1
z%FI6y&BH%l;=~mtZp(a)_(wYMp0!9k?eY#UJER`RWH;3!kE>4+4bd-z-#d~<54~vn
zu!OB%Rl{NM2>jGr1*coHaNE8GO|GfKGI=&Ucb&rYsowZmbPW%8oj_uUC+tt1l-<o^
z1nqwg{~AjiRkGsq8>+O;j-%dL@Lstv;x}vK?olfqym}c?^Cc^<5sr^7Tl4i(Wn3s(
zjcSK{q!+)Dj$$Svf|cMipc(#6%H@y`qd4`+aH?KwMUMr>>~ue#;m0=NvGpf}ly&8p
ziOCEv8^E*A{J4H$6f++FhL1`Ytgh?Mz(PX~9#Tg0#BgDSHbbGN2O^DhP%8I{0i6e<
zd$T0jM<$^E14j<a6Gv$K5$xZgEq^6+#D)#(oOe$fE|JqvbY~e}n|BeW+(6oX(t&3e
zOZM<Ipr4l(F02{MKBl{o-p2#U!!zkJT+Rs#W!{kc9m$pF(fQ>gIJ9v<%W1u7@?ahg
zF4Pbf$bGa(8bkJZi@LOG*fohqPoJakziWp5RkCmH*PVqyhjH>=3UpS=`2lCpcU~&q
z{q&&53i0mRX3{D191aVYIsLCW`*-U^`?nEHyLSg`w`lNK^%M9k7k>307Zxe!vRC|K
zR4Puy6+L^HlwQETlHv4t5x|BiT6D3u$B3!ISgcpzus&(RN^izRu0K%nXgzv;b3;dm
ze8yWywrp;DmY%RflYRl{_eZ<}8|}H~mGJcXmr<k0945KH(KKj2dYYMI-GPn>KBfoN
zfElO{2^J=OU-tI*;1{z3PB|(r9Zz8crWztRRTF&@#OaqFf`rk{(Aw6KUz$0y|H-Qu
zxio>EsdilcvI(rdXRuQC#szP+u=1rcH}?L53tve$Crj~8+tO02D=P+DFlc^jqy-#D
z=;^*psZ~YFPH!&jk_mGa@fEMD$EUY}eBRF!O)Lwz<-!U)TicSA55)m-^dC-jZO){2
zcD&S5nfG^@(Z|}3S5Gg(h+pTieUu{R|J{VYZ&u(<UkBK2_LN<w0t^ldW9;A_@p&r1
zChH5Hozmu#J!Z^t)W_7utw`MP7-j?J;%8g{A`fbF@{?q0Tz@P3tDbVUHN`8Rvv3YE
z=i_hXoOVSIiDv4;_?m>Bi=0t&b25??H)6hYw{I@e!KIBau;hFk)^s(7*`g`fE7$t6
zRSlS8+eF-=KX7r}S;Q<3!KvS^*zH_4bKSD=x|=WF%=O@c^=0&#d;kaY{)1hN1|FB1
za??=hV`@bs*|0xjM|QyEBsoXqG{K#vPE0#}Mp*d{JR9Gf!)8bD>Z_RuNS2kch92HJ
z6yn>b59st$c!52uppv7CheIX`le-Eb{w_2ryo0!@3!x{^m7jBkx%Yko);()P!22c$
zs_1}_XhqBvU&e-&$vkvy0u01wb?0F!yr|Bcwvo{2GaP$2PQa@8t@5*H;`re(77nvx
zz)s<<X8VXQ!WX}*X5+_=DLCqFM4y0HXg4y0BOAPhjX55MZwAn$M*_QUZ;oT11~KbR
zXGWWw;N_dqsCO05PIGM}CJVDDxri#eTEMA98yAcQalwWV4zaM|MaOKU|MNp(Qa(!C
z_Tu}81x)K~j*gc%BJ|c?U~(v&0)k-FR`Rj&&hlPo&j)$pIh&V-ed86lwX;2sPn`#=
zW*hKz&NvJ|avob6Rgrwp77xz$<cOWAd~>V=s^3UAs-F`=!&IR4IUN_@FN3?1Es|Cn
z@T~A6zF*Gbw^(zu>>&9kCqtahT7UxSAcjc(NzuZN+4A{2*w4gTqYU<){1!e3I`CJo
zmT038B3+IBIGS|}5ov+s>~f6%&lmY0gSc&hGED!(;p(5Qm^(-qhcRLFYp_SepuJGv
zszgJ}mW=-J4XbE}JwbC2c{>K`c!^i%&f)&Z9_S>`kbGgBT<AT3wijw(*;Jh~Wq*G<
zw}4q=M$tmfX6qbGk!+<*wHRaWdfgMtXZ416`)c{FbVc&<v$*hcGhBzYWqg}-d@0jn
z@M;g-+9x|sIrmS<)MCZZmRz|mOk5IE5x8qRs=wOMd0PVZR^NdwH=yF|Ak05&$BR)(
z%-a^kx%tM>@Cjnd8h=DA^QPMARTwATSl9c@@u0w(aV1Bgx>z4oNos8IxhcI>hfDUo
z2Agl|!6j7gdwH37(d;B*muTRI3USU)mqtgFQS?EB!8gvpT%Hf#v(IDYyuLEO(C2N(
zWw4(;1JySl;EsN0EVoMG-V$@nEGy#rL??cVZHn!09-+s#F4BWemaefWx^=SUSjP;y
zZ&hdUi%|?wZ^OyCzVuZOqTz;r@=hE;7>wc03$qX~Ko7g;t%OGPPN;Pmjx&zKSu?gf
zlLso$u&XXU+KS)nsN|{|g%z{qC+=2eq1383AHDd61NG_5N#21UU)>lqs{zyHjQ(NZ
zQQSJQ0qGsm5Th~<yOqUf_Ny85YlX42Nt+SU`*jL*M$9lnnJpS)<uebi>9Q0zG_{fP
zpfw-mMRCzBGuA&$W~UlkDmf_9)D_q>*NnD73H-I?KTI_Urf$m*2xzhe`6}LAye)&J
zd0TL!xeo{U>*KX|BL4cXLgGvl_G>v6A=ZoWMmWBIBp)(p_c3gGuLO$^DKuD`%6#ik
z__&RM!ju4-?w<vfdqL>!+KjVgS9d&HJY^Ja*7_S*sN{)+x5?bScnGam4CYx2@!z?3
z=ZF?fu<OlN7)xg~tFj6a?=5j|bPps+w=rpgc;r-Wq4K^b47y};SI-)l2BzZvP8GJ0
zUi9TDZMpPn7_!r(`zW8;#6NOR8fZ*~QJN_I1uW<(Jn8ilVN+8|ot7Th{k<cjgkKfA
zPlJa&fiC|4P%PcLWJP`a3$>^2d#R}WvcQTjk5RAegQj`k5%R!*oi-`MU_%uOb<X15
z-yw)KNJOZ*1y`ioNKf4dJ3KTP)^0is&XwZv%tKfjkw8}iH`Ye?qiT>oXRV$J6EjOH
z{ZYXD#Sc(oBj?!R$#^<$3zDt|2tT+34SYsm^3B_FA2C8JjpsO)_ztTYd@y2hclb4R
zLRPUcg3GcnxWGf0MB>x(1N%J~3-{6cas90xf8-=`ZJjkkN9bTl`>|NN?K^a(r}OU4
zc(^=hz~`6_n3gp}uKU6K9an>r207@-+2;GkYgn{$HKv^%%nsl68R$WI1~<dVa7`XO
zI|c(U_M)Y@nVhzU(XuRyGlX%rVM{+0)(*m|fkBx1sXI(2Z-w2eU+6dg0%8nhzgp}B
z4`Fu?dLTK~=VS0h+_B3_qX}1c&a5&({>EtjY-<9|&AI$h*#yChpCfHq8@dh%MdQO-
zILw#MQU@iOjsC@f>PUF}v}5NWd2g&U=6}0vnKjgg8ZqK}?HfbGHha)s?imiwKhSU6
z1LSYph3T5+*wH@$&;D$~t*UB_-Ea~<H?^7eR}XOouGDKI`O<6Z+|cGSww&+4#pkTh
zr<EV)|MJ1LO`-If?LfsR<8etiFJHWUk#=_!*PrUh!ozX~h*74$ynjSGeM41~c-BtW
z;bMhz_>s8|8r^>&uYWT4HpC*}STAOKnc~8udx*a8iVjVzxOCDlJZRg4HJ^{->nCS+
zzucX!JA6>nqXP|d7ou;=9^yJ*hoEaSQ7g}o+VUIdoYj<P`dYF5O)LI<B$)-DO&BOW
z#uEX;miyKYdxzAb>~eo-8bbJw;Z!;{7K0tT^3aQYxNoijw}aF0Dmxf4xvLNpo6Ya)
zvS0YDh^M+AP-IY!#eJ=r(a(uDUe7^}X>TU%vxVQ45Vq(ioL3c@>rEs2XF1}Z@E)JW
zG^5oC8>Wm`#?^MuF~-&o^K85DReB$;R<Om;rrp^2ksh40v#D$~6|p%B5j`Uw#XH{O
z;1BV}*Z2uz!3L`pG#UB90&~;mW4_W#>~Q;xqH;|*4ov5}9vxu!TNu5^JbBt618;;=
zzPC>1_imR_a6AWbOZ^d@*_;oziZk1~m+Wpg;Hq?~7ZwWlt6u!{x0=)Yk_qldN)P*f
z7TV<}VEwM{Tz(~={c5ApUQJcJ!-tUPsf+l-rFeVZlI}M<^7Wo>yz|?YC%@Vt?Rs;5
z%FpHA($2VF6$_nZA=LVi3Bx2)%*+6%X?o#nsuFtd$>s|4F0@@6%JVO}aDc}E#%&RP
z*?Bv>Kfervzb!zYC6X!c*^#S~jv?Q%J=>XBK;?D;rkXFul*jk6RLcrcdWMY5I|Vo2
zA$Xj$9V%BPHnDyX2RP=yZpj<;4wW-(P*a?6nFl?uFjRjr!mh8Hm>D*NAM}=?<7`JX
zxw98L4(H*_X<z)T4n**_A=K*Qgki}oq0lLT?aLDxSLcNHg<*(4AB?-T!|3Sx69Mvj
zc)e*e-mL4z_*5-y*;E7d$4MM~!jPLKuUS__47hL=X0hElPWT23CI!O6T9q{`RIzP!
zG(u*!#pPH%wDI@jI(tpptZIeO8aJx+w8uPg$E*?l!_k_>s4$j1bcsBRQf~^svw%rI
zGuc1a5akC0aJF2!sLkcPC(gPVSJL@w$`B;hj>L{V>Fky2j@hzH`PpwLj$g<@UCRjk
zRWHKG;cn8?>xl3(_Kg164=?Y&!-;RxV78zu^A|)gYw>$n+BQM1MG;rM(}D788@}8s
zdDmnIgnllfp_3#0<a|1MxxAmP3B+UJQ~gP5M5W~a9}2tsQ;Zfujl1%wbmGD<Cv&rn
z7J~iF`DWNH{P70Hwe8E674oiG--<bQhP2E41;30XP<vp6Mm;BNbGrroIXy6-XD2>v
z_6lnI9jSA)HS)C-skSeYg<cPEH0J|8|It8-bhK^t{$X{Jc<)D)N>+TN%(5;a=fqls
z=C($#>~VK|u0}?yIQ09V&F|}!(Ql<(gGrKsmwZy2^I<Gq7|Oa`ZYZ5+#m^2$g(Wr>
z&abvY$>Ad$Ub(UAh!W1(2l0c<_O{AS`0SQ2R^9sq>o0j&;bMrNrqi(efF6IFIAdAV
z6a>oLCH!|g94uP^m;M2Kvmlo_Mu}8VPej~`?r75IHtu{+=c4gbP&p)lv$`dryj~y6
zlkC_SEbPbAH!(<_^_Slz;`dDHRW8oLgW8)|ThI&F)eggTUMFm6+mDOyhI4L~8Tt?V
z2!YaYzr5EJ9B+W(k7z8LFAPHOojA6+2(}Tz>glS1FJD&TXsXP{ey)b|vSv86%L$R+
z0OKwic=A&R4i>L*x26Jpr2BRByf`#Gwji~s1=Kyugv}lghXhX=D5&tp(yw^-#uMw8
zrg8g`2IQ(3@XQk<#Cn;dQYDF%#{zh=G>ltjH)82+HJskkL0Eaw*f`-Av`&vg&x<w;
z4c6h`%Zjj&>v(0SG|ALybI%AvoY6jyhG(&;x9!0J^PI6~R~o%#SGjc70XPI~hk?5r
zt(p{}aF!-cE{_FgS+nk6CFU=4;>EQ(2<lOb11$&f(Dqn%aOi;b=d*Fdauc>c_hrK6
zAgr7p&1G}7Fmm4_B;|C(3#T-k|150rAnCMqiGou1VAkzFjE3C(SSmfb!&)`?9jis3
zO)>0j+!I@OH-X6%uvB3tI_$Ng+fXmKxRl|!jy@KKm(pk}nS19r0<Ku0mAQEAyScN}
z?=I|{YGRjOI0Mz=x$mSpq94myckm%3J651nx)}G5OaDjqeY;H8VvD=Hzw}c@_cVVx
z9sh%U3Xc)K`wg-_w5EQ>3dD7Zg|_M@EZsH@85>5UW@m!zqFb}Cjr@7x>MT2!MVI<W
zaX!hjJyi|sz81o4!zA<?*-_@8vX7nf6h{}_fxhhGf+m_FW9xNPD%?Y5<X^m~8jra-
z>e!c6j-}>`Xmn6Qx63j7k*0yC84+CS;ROu?Yktktgs<;%?D;Vkku@8^BtvX4*93#?
zxK+CI4ioR-zv=_n<9ZHCagoBCD<p6PZ6xP0SSgdY?i+GOc@CHANXFs#2ZRpJ<M^Ub
znrDv3?ooA^zuZvz>fNbRzY&uhz@n8Fn3ikF^6&Mi|1T0f%(P%AJcw~|N3kT-f#DCl
zIaGEu4lUG>@J|INdWvVkP=P$-i-nUdvBNft3S--%e}+9%YjkL!nkMh*`8;De8l?et
z{IFXQIW21urQe=qY3@`x?8M7%raV2X5i8_9dFG&1=%O`7e0-;Hvh!OM&raaezOJ}O
z3#MFA#N;M(&_-M&HxJigpi(Gv(>*w@Gy?kLTd>s#D>}{ALd`7)jy-R|!i8b<{Md`W
zI{#<Ybmf+2;<x`=js>oxIs3Uemd(Bgy$mDqe>Y;hsT=%VuHeWb@QR@!9*938uJ0af
zQd|v*?Q_w;JRbG8rq!vod|jg^|6mZ_h?i)$WSgE;2v?=k6qwFF2t7+PaZ#CIrVH_^
zL=|e!6lu3n1GcTzgkfw5>u0Co*U6eA9A4r>wJJ{EJcIuNe&AAc9YVvKV}W{0Y?iaa
z`LORutu&`nQY+39cAxLZP#Bgavg)D-Z>EVy&&rnn{N1_m!W0}!mvdWiGaBEqh0nuc
zSe0Hw<+D!If8xoGv&Ta>zz;e-6=8kO4GaBsk#jx~ujP4Z5M_a!mu*?tVl|MjhEqZM
zn3m&A^#$Slr(TaK&EH|H_GCQW62}0Yi5UKIIriZb4m_8P%LmD3p47qk5^ruvHKDO$
z9CaUe=h-_SQME;1c)!hf;GZcwOP)XI?h#m)i2pV*jBdNypgLHS(M5;wA+#9-QYOPB
zWga*|4f=QGei~hk=5=$hpq~XpR7c@%Mh9fH(8tbCUMTl<=9y$WrWOx?OZjMQ>=lH^
z)2(1G4E|nG?b+h$QRyypW}ov-acI2cLkpW?`Qr9a?EeAoo*8hmZ-PT#{TRDHpU0+4
zCiB8en7ib0&kr4FejbU!k<u%$wqX25$tU}$GU{v=o*kNkJ^Gq_uy%u7U-tO9N4!<a
z>db7LN*n3JJbkIcQ7xNNbF?$Y`Dwx+`8{HGW#a5}CB*DE!ppz9^w~WNdkQ*oL-kL*
z2o-1OiB2dI?(NjZKpr)<WJOtf-d}FV-qLG1xYCU6GQ@X%_b3`twBXe)jhn6${)v+R
zlY8@!V@WJu)RR-)x8dEKJ818q${oED;K+2kd}xQQ&ZBT~dMLxxG&uOkBSaNW#O_&U
z%>R7^|5i#y?0F<s_gIOI4R$<qsRye9+9EFc9OB!XB1~xsB1Wm;N`^C2gaunYVLW~&
z9Yoyd@fck(7RQQZe)_QmhmJUcBl>f2<Ut04Px+zWMg#m%^W^xKl1sm;L$j7j@H+7h
zQ`Ss^;izyvIi}4QC7pR`T`v11D`BJMDFmH%!N0HFX|&3T3hh%l|4JOKQmyc?o2ocg
z?K$;;J+(`wz^$t)mTs-VnO^77|CTZPKCZ#@A+vB)Zvd2qZC~Z7hK{-^4D-B?FW3GB
z7ifu_ezF@D8zu8<Cv_TTPnLV9I^VAUj><O17;s6S|9Y4(qB)`DtdH;J(bzCbfxDCa
z8T!lyU%&Oi)S7e1H1cL_ZVq!7{K2q?;_p7}$3Iq2pu5eFX4e9#@Yk6e?6r_N+n(_o
z#>k#L7XQ|kpkrkUe@lilxxI;GjGDoGxF*co+(wUS<ybgRewHJ$2m03oCMqhha=(e~
z{^t>!G7Tx(;$Iinz46>y{P~&6_jmJnDj=EhjgmP@QR3T`zLMcnLgwuY$dlcd<r4*A
zJ9wkdNMEeVxPT@<$m?AVnJ>&`*NLY18?T5nC&yt(bqFgS$+efPiAkURk&*6+m)YVI
zdn#N2a|`K4b;MQeYq;A=Jf=(Mqw@qe?lH;a!=SD-m;ZK5J9q53q(PJLXjW}d!0;(U
zsAedo+10LeEZ2vksSTsOq@OQL*21VFK5ivh*&DjTHok{mws}<En2E`iy6~G3&Z!r4
zSYfD%d5aC17u1E56Na!ufj>$Qw?d!Q(Tr1eLg{4#yvb-z)53YU@XP`z>cdTgRB>&m
z5$1>Ivu3CYF8_{a*`WK9b1OqPyQK(huZ#n0+|hP{yw5Kk51oCwY%sRsS$SXUr>?=y
zU37$BFDzSiU+(|pipG6U@jlLtro-$xv$uvY?A4I7RJuoJgumz_UdY)-43o_H(bIvn
zD0JoT`8#Aj*b#%fjX{FL1$?p_fsf~EWfuGxYga{3U(Fta<({5&Y8x(}9s~U!E(}cV
zLYHeg3^sj^!e{aHzIqn|6=%Qcrkq)4jfR-EEWIc@>-qnJM>&S`slsB+k5}Rew;Vn;
zGZS`#Dzz(9aWJwSS6&F?t8!tW#01c3OJ98dPZf<TY_aWWDMp^2k3lP9n7r4V-<s=C
z-}oj%HU)9-f@=u%k!x{JBv;5gZIkAAQ1i4BPR~^EU#AG>`QC%_RSl^2bLZ&S=IG^r
z3Ue(Mp|^C(8?*~>RI-4>Di$K(nGt<Q4wrqlF+Q1^a*?SvQ)<MK<|8?}#mO-4b_{Vv
z8nim8ilmL=LEXI*9xcT6)^9DA84crzjy;iHZq190vrz1;%zLp>?ER`cm-cQ&hke55
zd)=N(w@aq7+hklCwiCO4j>D4e+VI?LhoYPWy2u`CQl38Eb(j9z`9>^S5sCh*hG6o3
za@D03k`+@yT7v9@$7+a2UC#UQ3V7*c$b`=_i^-SW=2BHSN=MSV?-QgwaN*tWmbfoj
z%|6}Q^ZfM)wC*wp9!FHs{{-;$RWwwpG}+)8$k*ZZC^R<Y)q8_*qQy9@7G~Y=KrIIJ
z*JVKb4Wx`zfz!-EXls23KYpuonxi>~xC(zHWFi9VYjJhHGD3t^Xf-^X?H$+Rao%8d
zRSKlj!Deh_-5H0}Cu5H0EW8`6!-&$Uh`jR#4W`00UFg7Oi<-0Euq)=I6${h0m9Q(-
zF>JRUN)`xHw%dP*o%az5Edc%M4osP$#hm;Y7WZ<(3HMvjOZ<eQXcaaNXu?l=?eX17
z6YiEe%w1+GdjMk?3X5-(!frt3eW#b&^02n>c>AYd)B1C8@|}Um+)RGaP(q-JHd==@
z;zz+~988jTRNVk>zS#r+Wvk+5>I`g=d*X+wDae-`=#|I&@U>hQX}aq$?A>E{X)i|a
zVpa5N=ZZ4dSY*rnq~>`vrrhYl{7r3Hwkr~r$ARB(_n_yD=1}~lf>mo)p}Nq5Up^%<
zy7heAxzz@FY3J}O{{T8|U4(Y8NAS6DZxjqP5H_|RXFW{dksy3c-CJ1iEFE8+Y}(Wx
z#9hh#c)nPKbbm7(O*7!hx@ugyp2M!W<oLuTXd}Cjb1DfuzWEda=LYa^@>H1BNKU7y
z1>Q~>4@V_sL@T;+TB9;bpRdDjVLByAm+IhWBYcwIOYpV?zA?H3Z}IZ1%+uhCA!XQn
z!;d}-f1ysn3GHwB^NcvubV859!}}ggrk}wC7kTGW`3v{<zH-b_MC9lU>WoWZp19w8
z%-kbg2gy=b>ml1v9JU9}NG|UMVwUQ`td%_TdPeZxA43egJPz|;#xnbJYpivYKHSRy
zq|cm<PcypkZKqS1H%^zPrZMdMRQgn@!fv&FiH;Gz$Qv{SU*zw<$wYFq+jJ0DqQ?Ti
z^|+Zh4xx=*sXC>*uulWw5Op3Znge)bj}3R1g&_Zy8`o*%;lTi9+~1}I#kve=th2-Y
zz*w|u(HZ`#fxNj-9p{{z(0*1tk2E|*(xK+?iCT)IkH5fHyOZQ@g>M>P3#Iv6q;IE>
zRo+TmyI2V`W{JZs#uQqU=gR%vPj(c-HS+4h{=N!O8r=pbM_J&p{Ml3c$FQKod>s5B
ze$Lj5Vb;%;kA>CnOYJsT(hC><IWvE(3cX(J#<Lc|=z8-t!e7aJrtu>JGZ&*zu5hob
zji@U--dWvz`9(4-SKAL_!Qvp^y&i+fk~NwWc?4(oXyWaYr#S50l2?{lVzF5RjyIaa
z;C)|qO0=d+p${I=72i*3pzX|M*pjQmoYjpm=pV=1S5#<ZYKT|2-lA0N6Iwa?!#t%I
zDmJGv<cSiN-t5V){VEZ%vLmX#=HYQr4r>p|^XGE}*Y5m@>OF7KIa|)WMe$Txp@-GV
z!uc?7#*J0ZP>fCHz9k(b7oLmCgJ<wyOIxO0Z$#h#>4c^nN6q=YSg>^?I*n7u5Y2<w
z9d;1TC9Sx^9W-7N%Cy32<h$B2M`p!dgSs)#!j7Tlo|11+W&iVJtI1LLBj<PhuPbm!
zR|n@Rl~8TJ72iiBAUj&-36jI!=RXB&p3TR1yKZcEG>#YERp42@B_~S<ZGLoL=-zJ1
z%BVAN7k9DGbSu`>ug9osv#{)>GB?RHS0k?;DjRn|-KZJ-mPsal!2$HLTZfcazfk&A
zot4LD!Zv&~ie&Ddc&N3^>CN!8TwD%!Jn`6XAJ$!(Eq&e~ejMe?D_whHkEhJKr#fJo
z>tJsCVvcD^9jMbdg44PMqsqaDWkdT?=d3WAyM%Ls)_%NPauPp2oQBh#2-L}}@7vrQ
zoKUL8J$-jt=Vl>5x?q9bm3g?wIozD*!mjDpVRs`37YnNK!YhkbAH=JB!xUXBL)d?B
z2ro|b!i$l?P#L)$yZ30Jc9lOD#3~@c%MwG!nb9U%`qc|Hkw4#$ljN*!E&orAXa_nO
zr@}xy9_6o_B1TV~alvU^-$svj-|Jz*AzL2YxdWFx!l=3@h*g(6prY|0_It0v--3-;
zQ+XB3ZRDLrGm^hX-G}MLt<dTH7MG0I!{mA|Zq-sjRM){g{mC5nc0GbY=07Y<k$3vR
z>Kr#$7mLb=aN&!dP)ctvTrNA@oMMTI_9}dQDV8N+Kx!vzv`II@<I}Nh{i6lGFSdkd
zW)U-8H{xq}AD(T}n(gzI(CTS7)Xnpwsk@=P>qQD%YY{ZwlUL`+PG?yU8YCyPrDhi@
zjCREMNL{7}EQkIE4ZMFPd$r3Qcz3=qc)V0`J3WCBX20>Jw=ZvBbHkXXEiht&ACv26
z;m)Bp{MSiccpwT`v#KNJ44H$iqyM6bnRLxpyvOdpM$AhHW7B``bhguCm1G}hixb4$
ztPFk~lxSRH04;qDx_(tb9ecCH(u$JY!vt^PCx-T-&2B3yEL6qiC>_ZRN}j9GoaWut
zpmY5)n!20wiMUFd+qcD|XSQ^?_X8S{da(3ajeVn&nEg1DRgb#k>g5TDEz#lhqzL*R
z`h~U2_Mz1gHO7le<Wo~6WVf$G+*5b_9^Mo);`?As-b{QQvJj)JB=b6|GmOqwAR{mr
z5r?(7c0(+U4m_9ZufP0zBQ$jLV*Aa%ab|%RqHJss@4OUxs*;I%))zNcw5QpipBN)&
z&iCmG7?5=v6N@#WK71Fp?T@FO<ctnw8zI_8+<J;CXlHv7$5k`1x?wgdo_k=WfqYg&
zT8ZDTEu#zMULB!?H7`!!npYdfbxWX?>>k!C>k0>9D^lMbM%)}>CGS#U#DN!R8ov?A
z=hZQ=zc$}*?n;YW8W=Z01M?(19;H`-52_uw{ErFWh;yvMdjSR?(187S8#FDr1drL~
z)EK@OSGz@Wy?G=z%zTQcg}tzT(nJ(0xzlxs0hJam6PD0f9JG^t-p=U=@K@mvac3U;
zS&f9`RhZl@ACYpVQd?rk|2E5<=;HuvTpq+GZ}s`!b2e5;M!Md;15GoX`O+>6yB(5{
zHe(sasu$zt;CPvx>q4P#3iSpXBcS7bB(Ds>%xP_~;cYM+g=@ZWyFJG~$zr;&KN6-L
z$K8GNk@qPRUst(G=0=$>y6DsWq#2SAS>o`<SvdR6h1<7`fyfx}d&wBN#wX!;d*Ryz
z{DoJQ9-c3&!LQy!ar{mt7O$F%?VlE5;t8@i%Z+>gbLEtiM$j5l4E^S@Xm&FcyYBgM
zK#PUYoaBja(lKt6lELBiskqcVica#G#%x`Ix{&7N-Jh73uZ5GEoiMbj6^`^!gJ+x*
zO<bdSx+s<|v;8?|WEZ~MCw|5@Y8a$45c**mjIekPqdjV{lF!mhVG6vT`OxXE53DkK
z@cM)6SRYk|q~W1dt5D^p-up3dcM4zabH{n<6bD{X#FQkNv-(FOc&Qnim`ukucOAUn
z{t=fG_h8u#$@D*LE}0Zz2+wdve*GCRyD$5$YDB-emN+}s7*Ef4!<!c7<o959`D(&j
z(%H09ScZ>P85})GoZ}<avDHYp$RRB`bnp(8m$yXERA=-Ey@m79&di9rjGmYKQ^E#B
zfg#g9?fF63ioKT@V|q+I%&L@dKy?Zu_l~Ca-;vz4>#FRVz<sAD^1ZyX&Nw)UZwy*9
zt?f#vS#+ny+(M=w&7^z4X`Gm5$NJndUP*1vapT?K`OIEe{Pr|)o`hiOk8UWInNj<$
zTzg~?Q*~w{Y=;efHXp|W**V{Fs$}?`P+VN>gue?UTY6|TJw0Dx!gdWle42wVo-!kr
zen-t0;UY}Ei~v39k#6Y11NUrs`DITAOqq?-AE#p8EggDBq>BHk97%nf@_fB6l!PH1
z5gpAIPb@gjH5tEeWwUha3mos^&fACdQ0;XR!)_62R>v^nQyPDCJr0KtL-FxKIrcAT
z%`%HzI`#6#=!S1N9=!%5X9|<yr7H59TXWpSL0q`VfX$9f!Q?dIE>D&2x3+M~U6OFC
zaWf`wufXl@%6Omo0x3s}8Nb<`vxbf5TdgyQYBPzhv&XUgXe=Tv?Ab-Q2^~fUvdm-y
zG$&TW$ZagI+}5Q{U<%vh39GcMglCTo;n>lmS*Igiw#;nKyfTzqN4s;?uFpuxX^S<9
z<LG-cgk>caXgntw=yo#@b*39bGs`)%)`p6bKg*n0M&Gb}TAyi+3WI?>eqWKP)0{b^
z$pH3R6u?uJ?dVyR$x-qg+cs(jx?BgFUw(;^h5>jS--T!DGq_iO4D(vY^SrR>o>!OA
zVO|Y_GD_H>whK?c4MN15u{hVvnV-uQxq9{}o)AA>QA9I1UQFlSd_|5?(~|4J54!gx
z7v?vbG2gtIQCE(v@OVtVWsA%9`iNP*3!e`L@~w0~)PD`<P&w}&I{+RQUzPHL7JQO0
zo=dXoWe(H}D{7zN-zjkreAcFeRtl${OJ}X}1iqL2UjEo~2yq!hkD@v_tz3emPlvKx
z&N}9v$>`>9!Q>W22>sEI3e3Qr;Z>+Q*q6`LpTIXgi5usR;b-}dojdskA1)};J^Q(I
z0kWBNshD#Y7qDH)Kzve2fNoVHKI@HRPDwbOo|VzmBM%q<E5^2fY&_ALg+n$o@uGMq
z?Ao^A4@Eol3K=4yjz{qAb_R2&T5(;yHvfI7$M+2_uy>2}?~Q;LiRm2dV}hWsO>wJ7
z7mU314&{q_GjT*E&ghOodb=JpT@}soEv{qGs-EmS%>;Q}-=oMun=uz-i0~@x>>q{1
z<ti8;`D@qi+2Ygg%q|Zqk^9dcZNpb#lWG|(JiORUHJZb&hvL{-b;frX$IpsU@@}yf
z_1AMSOm`d=vt5{VI3DAll`!|R8Lj_S(xV}YE5Av`^k6sY`1hkhZ7QD)|AWOX%NaFV
zjRs3<aXWdUc#OvJ=&x>6aaf9KMPcHew&mNc6<jyOnUA}cv8sa~)PA+WyhZM85Mu4x
z-4mdBtThjP>;n&zN?v_15ei;8DDFCrajg=0y50qYmmES-P#IC9PN)60SUkc@`j#rJ
zP#%ctqF|19il+5@Uo;<>!IL#r7=L&SM_YYCuL}iK`<=~ce^Thb&W1aS+T!K+r$~F>
z8C}XoQ^jv2_N$G?pYxK%H*!Pd_iHG1SK`jX)iAy~gtt~p=IM?Rrd9?su+tLhQmDh#
zx*9jklIXLuIe)$Dg9Y0Aq}vIuo1!G$z)d(ac`|=^pU3b}VSSyFJ^54z-Wi&WkOEc7
z0v&>%-88tlcH%v6Z<<^ggP{Ru5kJ3_FYoMz!8|j1x2mRBasUH;RhU$F2=OVY^vM}Y
zwb&A7tA54Z-|O&3d~&H@SL0@5C@&kkv6tUCzHjKmUIsp#;a-m1Hz7>cT#8uDD4spo
zhyPt}L2K`1%<N$$=Z{fvo?5^>$=+KEe|Kn2knn8VVMIg-R{U6thhbOHRPuy%YhBU#
z$Uxao_2%?e-z1kt>dmQ_Eaq788@6Q27qQrSIg^Kz>k#-)AMvxrrInO|Ju3a-yh<2z
zcJ{20Ja)!%aQB>i4BGJkS`&=qIcf_NtHCJrwnXOz&+)t89jxoE$*775yq5PX+lW}=
zhV-3(b;GJgE9&1f!`<ywFg|<<>%I(U=Z9li_To79c1q^5^Qml`H;FHYOr+tSERN9c
zNZZAQ`1)ov9|Z<*ufcG%dE-e7*>M%z&V=RoGFIz{Lf6MnuE|?)IO57blB-*F;Uu(2
zmhwuN9=~b+!K<SKaQ^imjuNir9^n%P4{0u3yH@D9RXQ@_q?>g+kWZDPs1^PQjT_o?
zckC5R`X5K<9nWR|w(&~YAv;?kp$H*4j%=k+k|-mTm6TE;iAa)`_9X4Cach_M-dlU`
zUE1}Wzvu7!^}4ICuj~4J&ht3l??K7zA8P^A2TsiV9m@_~Q*dR)7%ps-j>ZdPsLs@9
zruz`O)kPv?nDj_WgekJXo5LO#Gbp74o7`OtlQU)*_52Ik%xuky+Gp^0YtMG?ED^FM
zkXOT^`OIP-KC=!}lMdnXz`<PRk&82N<!G`a9d2a-43oR2<~u!1ojZ&dW{C@Grh+(F
zg+Y0GBHYtTG3mDhf?`x<PLw4jWm~2Woy_v1g~G6{VC&dLm^orOen0Qd<gt@@U}<Oh
zET*#l_E?4#zsK&)WjyOE?0M_AIFWA+tNTWb87_XfjT3qAwIw}GOE9)lSc$#M824Ft
zQtf1?lO_W~-&8v5IH6&wuyLNe7T>cEP51e8@}^=M6bys+$6lBo;DCwS+&Lp|DjGU=
zM(<SVK@B#P-`^pe_N5Ibotcdiv;PpBVTs^IB0b-dW30!q^6FW6Hw5#vZ4~<b=ga$!
z^ANGTKhGW!4nXKE1oSY%$9D=$bdIO@jYX)OB`nENCU|bvnqC9V_*3#fbsKx*S+hv`
z4_1L`r=hqmJyngw4#>Z>44(&CaYyqEw*6=c6W>Qr9<~*4WX5|}yu-<D6S#7T5?xoV
z#{T9xSbXI=es=MoY48U5eQr(d{K@R8KSO$9U0@L@XZIfB>s=9ph-Qb--D?T1jDC$N
zeHE$QqdVRElptFCi}`o^Q0+u4$81;)``vQSjCG{VrFiyfGllP8Ml#T^1_?(?a6)!5
zDwRuMutDB|-N*9$q0Ts(QH`&vqnOn?m^(%-$Cuw5p{+I#Dd%ODI?{*x0<G~zx-kg{
z^5l2Hn~IyWSe)KN_SrsI@z{VBzNgXqQwFE3-HYkZ2V-4PBnzARFxJtZ^X~MOOq(0C
z&T8PglLbc39tV?WhDd%;C4Z*)P`sjrHD8U2sn=n9MS8*)m2l2#I&NQggVUW-wA{HA
z+3%Wh^U4V5#d`78IRmCHi^Z9ERS|Txf>VAR>=H{b>uVP%6wQKdnfN}cdc)IOiDtV;
za*RPWv|PJ#%uWxsIS|j&`dh@y9?8C!$McdfIhFpM#ulX#X3X4(Ei0p#xvG>Ot2@*2
zmn&1`oR_IIk@FTDM)Q<TJn2^??B$U-6J^a|9MAj$MMg-L?M3hH(65T*jGM)*lZRG+
z;Vwjr^C`I@16Kc1=o;k0F2@q+)iw<46^2UBb07~b9?Ju|y25WV2eoamw6TrMqkUj?
zcqAtbc!<A)vS@0587kRJu&zxPXx(yy%Lln@b!knj6~!<*=gog%!t0!W2mcj&ahygD
zciruc;{KI5v8N|a`%7M0Q46_`y;!3@1BYGZ{@xJFZpWJPWK$)ms5D}U-YPVkKZw2m
zj$^1pOK44Pj{PfzF<#ONBihzr*2(F3;o2E>?(<OT8o@o$VO+9&J>qI|=wDudly6gc
z@Y+<)JzB;w{|Q?!IfO>jTv*#_HlAB)aAw~owB0Yyo%_8x;(h`BH6B5wTQVJ5bmA-F
zHZ`m7jPq^3V7mPslzth(h@Vb8+|r!YcY-<5ax_mi3csQuke!N_<E4EuQo4$heXAw!
zCkOM0$#|MXTT{QuXl6S`B5A!TI$l`{w~a}7&jaY3JBl9GKB%m)#^KOUP?G0mK#PUY
zHB#pUMd8)g$)0I{950PGWT=@FdOmG~-fi!}q4petS1m*Qrk*@>PLBo2dR+Sa7nEBk
z;p<joMvqk&UZ?{175qcR?74Wk?SgQ$w_=BrjW`?qP#^1oU!5JXtX%*@?cbqCd>+pT
z$439-G@N|25S41e8El*fE!p!X2YiO!c30k8-IfVoCUVs(R~A^$L%L56sy7}*uk-?h
ztIx;L6$=nCtsFH&?qKlKc$x){<on|jIbq=x9yi(#pYs;tos@n@#B%%`7R3*G{}4C&
zEN<NMXUc?PhR46a%i?6XeoNut72+P8KadK2b$B|*2UGhtAohT;Yzm|KSUrxfI`)$1
zt`^TO7RSnQXDZ(~4&RD1@Ux1dj-@sI>)p6E%!g-smkJHRne~QWu%KSt;?=rr_FHB)
zSKs1EKXa-)e~Xu0N3*EO1>7`^MA#K$20fE}*>Ek~(^`bY)AlGC(3eXlOD-m<1&!ut
z;cT19sL6YcSM^<a`bcNqZE+Mi{*j#gL>JB>BhVPOUA*_}pp&7%4ME~DY_<|t1{%Wi
zmI^Kp>A}xWM#5y~9E9{UW6X#y{9>lbLb=cFU+@EOFO>4@QUe-iKZM)GxmeSGD2k7?
zW$M^nD3`scZZq*&KYR!chhI29_$0Eoi0j621Rib<W58SC%-l8S%ZgIiOqSeFeke<8
z%b?y6iHiGP*p^m@)AOW%R~N{?-+rR1TR*fFw#??IeP}6c%LwZPc>dar9^1sz>(GZI
zI_<#sOYgDc-3Cy(9fs;ZMAQ5jcHKN5$A1~&b)PDPB(9R)o*p7E$us$D67)aTVEi$C
zG#<=?)>m7+DVBGixE#)mGNPk%TQuxS64tXf8W#?s_pc|Il3l<WV|$EfIvK0(cw)@Y
z*N8V*kM<YiIa6{DNws>saI`rR<h)_{&5615-(p0?a0VXgh{Z32PyAR%{5I-{eqDyC
z(_>KBzbjm3kK{|`NnCaB3w-ZC!aXZX26yNSFZC4MJ^uwRH`5sBlF$9=;&p8I2N}8}
z&~Be2?^ntD%qLa+5wUzUbtv~b4&p=zb7qMeKhK{aIr#|MYF9u(+k#3J+5D5-p1~Q5
zVG-2>8O^MOC-DvzQ!;2YN{`c&9OyQ97rf#l*^t(nTe}*g&$t3UX(*wtPPlN|Qs5xm
zv~P<}!FfKpR?eCRw|g<&q!>MYBiMAA5BFYbhfWjhWPjKdmpj%X{-F*tgz=!Y%L&gl
z9l5RIC1MKY9Cbw;BQk$_wlq%MnFClbv#Y%GW}yGD)##C`$?U*3_}SNnZDuH8*S;c*
zXr9PplEu5%co3G7M|Swofi+!id9<?^>d%>=z*U9w>*qo@)1T}1Jiwc$zI4011f4=v
z`7U84x(*vHJrTL@dwH|vpXsoR&0>pQtvI8sm~*r;#rcuPq1Kx8R^KH1%p(47E1&W9
z0X$pZiLWz+o1Pnmh+h^o7bcS;XX0bVd7K%z7*1EtLBH2OEJ_kz_wf!`x_%s0JSAT(
z{Vwmav9z%mAxsK!Ryq}PlQ_hm^)2AVd><U>Gl+^8#Syv4i;ovHBHdbz(Je~xG;|J3
z6cX@vsB~OP+DT9BI!=x=V)n|(@S46$GAS()wWTL7EH^=Bi#2G`*n=G=bcE~WPH69#
ziK9B=yAM!fO3-lbySE!7$F-C@s&J1IDiPr1&$#BhFgB|ht?#dwU1ToTjk*Wz(em$K
zuFuJm#rk;W2^3Pnh65S=<*k6@Zo0VkxeH!wilL3mDcBAg%r#MW5%IZ{9nRf{!l?`n
z%@v1z-b8xo3iG8>_K;ofpd`XuKF3MipI?lXEvE36bU=0{1k-VI5LdVQgNbz>Trewx
zZyp?mabPCrC{|1MWi%~cnsH*SBVvb-W=(_*bl=|*?ppzlzj=$sk&;0c-|@5bVVtGj
zfpNPgu)xtqK9h6U-E%e?Tj|m7oHA>kh|eq}i9=phN$$^$|IUm1W3D&GN9XZV;x&xd
z(nr#zcC;9#$BD(g=sKq*`c2sl&%Zf*dMkl{ew>G9L?v!-Gse`wFG&A!3qLPNPxg8Q
z9^2ca+|3YfSGrO8tvhd&9zqjmH(1TJ<l@*Wd^S_%x#`YGSlWcg-E)K|Xv(8K?&I8z
zpU{fVN54Z?aO-zGt-_4>JoOz`I%?22#EgY&`(S5M3Fqcc;l1s}{2lMhW>tG|uzC#+
z&J4hh<wcwn)>rsj>u`L{9y}X;755HCA@EEfoTH1F_g->g*Ctb=UoQ?wD5J{sWl&sL
zj9&3JbZl_u{2S5{J`61R(4A`IedulwLF4Ks9OvPNM^l`*Q7aURy`^jO!=4S#m*bqx
zNG5*u#)}wl-fc4;uU~tz`9UA1=^OJyWhWTbM)LA4W3-W<!GrEOut|D~4UI;q^9ZI+
zs0)WgC9?O|mhjPijp~bi*xY;sMkFL5Z>b{xO2@S@?>>TeNgkBlpd&fkW@;Poe9LNt
zc+G-uqa*HbPGpv^1DYO?IrN4hOqHHp6BT8QFKWdPO@cW;E1aDg$HBk7urTta`!&-F
zRdXHKqrd`39!l4!tOI|K&%$MmD9#$xil&oucvL-u$1WVg^Wnir$kpKk^%9;jE2MfK
zVnT&<8z$?r)ipOfOtE8TK?MRjO(Oow<CwquVD37M-_}KOc#lLX=NQoR=s;fm&w-ll
z`*G>(_N*K<l-o|t#;HAvv9)u1VbAxVRrg?gIxv_^75lS;Nw9c=+c9W|F@Lu@iR4j#
z@S&*|r-gK6iD@vl9dF6d)pEZXJzCh$hCI5~P<*A5N!IeDqqqge?yZt8=5;)hJ#Dl@
zBCYHb(RN{|IBHt+sJA^HOv+%chdMK+nnL}UbOHQyn7zr6w)%C@jM{*;X9nPzTRe=n
z2BXibVO*;wY=w_1Xt<Wf4fhWq{^<o=)ivOP1#@7wZY6?GnqmAlMW)Awa$|`qR6a>A
zS>Zo4iF080zrqHX)eD(T6S=<eDmI-iVVi<nnjWseg-$n+dNu~*67p$sJAm^-^Vqto
z2wB309%?g%AH1!Zd01G&-zK41tNvIqFOBC~ToZR{A)gKGj0riB>?reCU+EMY`m59B
zcP4DENcQqeI&bCqGikXI_qxm7>Pu73uNld?&bA2jZx5AYZd^1pLR<)5{EJ(tdzFYW
zos?Mc`63*&ItXLj9Is5o-_+t9)|+SukMR&9u7Ae(gy9IEFKnLq{bA=i9+P^w@O6X+
zJFYY0sbkOZU_=^S3~s`EXfZ6UoOtCz0e<f5fWtmAAKqMq^D`vpu-%DnvSZoVJeS9$
zbJVx18s8}OMCzWma8w)wZF2=o47-3uk>YWRScCL7N|^F0TRZ^9D0d&v0iVR7Ic?1U
zXW(5H8?nx86bcqMpz-fyJdCyC$=E^2nm3fw#^o^aZV{)=yo}~UQW^Rw46&{|Q7?JL
z>yHPa%_%1&_>|IoVpnR->yG+alX&FN9fY3p#83lwHmMs<+dO*&xqnC3+1IdVeQTQa
zwBf9lTEbPT!@SjT>@!J;UsT+Y*Ul8Xn@wP0R1E42dc$!{5;srpkI&+7zB%kTrj8jv
zZ%)ViB3<@saS$0kGOx<I4du*iz8PPNkapdf(=MN^HbR`T6AI<o6|**;&O;;zcdv-w
zBPH8hqKS<|`tq)KI(6?F2s=sgt#guiblx~d7zQ$PRtr2F)EVl#KB0A>Dn^g(!-9sU
zY?UFooyo4U<5`I3=XXQD&v{h2$-dg_JDR`H=Qa0Cd@wZQiX+YF@vIU#(>3XrAIX_9
znamwh%2C}H2pfuY6-IdeQD4mYHi>P%kK*YIe^FpCh&vYa=4bJDxN;&-Ou2zW&N=+=
z*=%tDC`ng3pR3PWq5r~MPEskBZuvCi{_4eK!|vRn7KL}KY+?6(7?=GMm+7D3xEH=0
zy%rnd;+auYlKFPQ_4ZIr{s)hrT{tYZ6rZI(bVZy~Khnl=|0!WvImqvq%=cF{OTmI#
zEjb@mV!Fu~4%6$59lzo+b;%W^WG=z!uGxH%y$YGrjD!hf!QDohd^n{w6;3$w=ErJ0
zs5>i+VJl%_-b7h@Te|hwgxdEP;r+`Jecpt!&+;w^%TD2E0}J*z+n)O$NY2sqHdNlO
z#~Y1k<lk;VGc6MwYuOCLzcpvEsTs@k{c&M=D2FNsGA`E}7fZ{SxZytx+%tqty9IIQ
zo+><A4aQF|<E`cQFnh%v4E?qheqMbTAZ-1#os!{kaOA83Ve%Xp3Z0GPdGes#Ju;24
zMcij|rys<P@dJ3Y%Vg%Aif8I?b6k6;%~qZA7!%ZnS1MBJ5mAD2r1MK3Z=8|ybz%Q#
zOwu35fxCuq)m(e3mRvyQnk>FuyB;_!eY=y2oOZ7VHO||z^2=fI$kkw-ynBC0mT&cU
zHJo143Wb`XjOcO$J67Dnga+v~-?C$5QZ|picIJQ664`ck3z`mY!B?Zl@>uC&#0vl9
zh`9>;`c&c42nSZ0I58&77y-7?9OjwMj>XMke_25~O9RkGXAqk&oP!T-gt45bf?sj6
zZ~LZBGy4`$&l|yii-xk9@o=^g=Fz*M3Do&+$9*f*s25TRn*ladk>2<raoc>F(jQto
zqL~xliT`?(prP(Ltm3ZW-NoZ*cv>hv-9fBbV#%h`?F<ZQ$-bH;ywPzSgVs1>qx@`^
zc&XC$-Dt*{PNJoX0pb=<!oUaN>}+n0Q)T@ySUd{n2bMzhZAU8A_u@)nIcDza#CvXS
zp_(3n|60ehRj~(pS%CSH#WDG_0+$~eqrs*ccP%5Bkn<G=Ha2)3p2`JE(s%bYg^Bu6
zeA=GHa>*yO+7m_F`G#0A!-fNd-|4(2n!0OVh;v%{(*+IC^&QMDveyePRfS%j1MJdT
z;q0L_KAAipitW#1=xZ%nC7nk1HYUjWCo>rpnQyIYEq+uTUR19TR+j;6DmqhZLK*ui
zA48b<&Q2wjas4hwj(F1^MyZ}y_NJK1n~p&HirgjQ3%PTgbl#Q9cvC%`F?YMb@3=0F
z*C}x1FnhfEz6xI?FIXy^PF>#+{<`af+y8BcaWlz(KMse^!dPw#F<`fh7Mwhx2_62%
z<BDV;zZ@9NcV8Si@Pq>*Ud#Vub2;+QCCjWij!vIt4l}n5FE94vu6|KyknGjU*JIf0
zWQ=4C=ApKa2HW%zKXy8x@8QY7Fbyib$mX)>&S(sFqm6A-9_=uk4yA=WdbcIU1?F;Y
zV;5=;8H73&EqH~;ptgS$Y92&MwW~Q>*S3VA(P#_}v*A&{5=;-Oz|#QX%jmYHc48II
z$Slu#@CD)a7()G3bL8)6g#-ISSTH{vUp6VhzL_(W{^nw5N*N<fOZY}Q+1b;^^Jns3
z;jzC(li?>ZV$)=Pu21F0Jq7YzS&!u3#dI#8%-pByyjCPE-_5afJ{*kuOM{th;K;kt
zCA1UordzX*xOLB-(O>W5`uqgyZgORlf3Fd8sx3BZx8NJulMe|`<k*k*VdFYVIBrFB
zb<E<o+kN=D*c<MleVI7#3wk$=VcsKind5|Wxp2hNh3#baN)zw8MPha2BAMrFVDAkb
zVJ%0pezqNrtXiVIbyI#{sz=}U|FFm3N%*4TtFIcw*!fnRJJx{fPMfp8o;hC23{)<f
z2wMFE&z(-d*<t{eg($*!N(q*|aEIdH*6=i1fLGUq>HWS9R-g6cqBjF^tDi9kOo^xG
z)=?;|GiT_(rpPIlzW>wKd~-q#XT*{AO}OhWQx@U<-f1{;V<@n)6P=VDQF<~T3pR**
z<>7AWE{(?H-)__oFQvLkUo1RO%DTOyxwK*^zQ=jPG&Tn&U#DYY<P1F8D$JqVv&7lg
z4)r;Qpk5S({pq35yz&;aHO6pFT0W-ioCde%J@`!_kuf@!D6x@QGYWWoeIXC*4rJJT
zajeM<sOFR(CZrVeO#il==KdEc>83cDZGqlRUgPt@So~9Xfk&I}!bMS-qM@<ex;>ja
zo>xMrbtD&PYQuY+CERY0#DaMN*!-k59g2>k=8YEo4pbmeLv}~W25_rC4o(YWFPmf7
zbbB!_R;i)WJbfH<Z^rGeU3vXu3mQ(f#=SnvB|Eeh!>guanPwtxtp0<Ps306Ja)3r|
zEKch5#}=37_@gxhA9n}v%vf(snehZWvz8)pUM6dj#LKMe%cdh_|90d9rk&i6^Os^V
z!RrNv<d<;bmjZUTOyh_i($lV;2?zau!q;!gn4gwNx5(p!RLOmuFTmD8<7m~SDPJ2U
z;d9GyzPn*YZJ$}FkoQ%coV}a2>BLNz5QLb@yJ_|xXdU_?tOFASuP8-m**vHpb!MlD
zU!k?(0J^p4$CXL#F=<O2)f&7x-awbB_HAKE$qg^Lhc-tw@$BPF^yn;og>+M{ecvD9
z+g0)BbO%(1%llPI(esQ_@T1j#cwXee*rDB+Cd{FQ`ZtAF8ijx*>Qt69=B7GVYHN1J
zy={k}ucwBmJ(i%d*pYVcn<IN&CHC&rWW{(T^m^6|iwBB>t;PY0v$bFtBYFH?;`uIh
z<?~0r41HC~KlOGjPqU=^*l{rY*qiUAgYBT0grS%F^903%6R*XB>^-nju#g$1A4}t1
zSo|@XN91?l!3CMud9~+x=@X`nj;4pO+J?+OjuFd0qSD|e`l}{kT}~RT%)4UzZV$=e
zhSEf*4w1dLp;<_Car6}P&H^=BZte(m@l&=M(4C=8=ELRj1@zl3-3aNA8z~l{=eiMW
z44MPA)&^L(auA)m2H?$cO$^$egvh0h&@fkLbiSXQMXuv!&*jKS8NhP}zI+r@A-(&#
z$Ph;4nA3jv`fU*7dyeJCq4wMw;lmHbNATcqFDlJeL~4~QUW6E;?_yzt7WPF}=^q?Z
zeTfk_-^g9`4GKCY;ERqQ=MD5i%(6hlct!9)vbd-HH{t58->^C({_h3CTDG;~gIBtI
zJ6GmvqvCkDW3t59Z)2=<8dtv%_QmzH_;Fw&pUNJjdb=;8YU6m~{&22wd5otoJ7f3A
zF6eOA8cw#>2yRm?PPT7Y@Sqgx1||IYa2kwSr}FCO47A9<EBSKS)9n$T-NQmACiRkj
z_)7FuJcNc(MBD{Ce4A(jyJ6?iO6?R(ZN<ZM%M+LE_v6eO2l!|1!QWnKj1LdSfgls6
z72kmMUM=Q2eL(2IS{(Qui%>@cC{OEziaV4$iaypJH^9x6<I&rFIowUsY52gKhmK{j
zC{~<FmjCl@w8q(ku5>yd2Y;7bsutTqyIee|;sq~48lRo~Uk3O+UMy+PnGGFSC0rY`
zcSCrpFa_l$t$0zs2On;UucL<^x4I7Dna~U_2si>mElbwi^I-Ei6L==i1TI~Dxa|0B
znUC~B#+S|L+f9*4YK1I{8pHl$a%iymG&bZ(-s@#3G;U|$X3QWMnnZE_3O_vcD4}a{
zC?5S4&WgDWI`t86g4%SLcbv$!Da)W{>Von2g}d#%3yHB|xR|pM7I$-T!o`l8#fxNj
z)rc#U`ttsUB(C~-9h3dlsL}Nb>^+=tNA5Jqxv|XrB=dTC7DtY%$A{Z1aLjix3giyc
z_K(bZZ)aifmEqVc{*8H`#0fjGQ2e7KSw71POA34QtfCjb1S&(ln=bnFH=^Z7E7%5g
z<-)N&xa@%|l0OE}azr-AYPP_ZCf5+Bq=(H(ak6`p`|7+Bn9Lo)x7$2;y}B3lt#n|c
zp~eHI`YeCk30oYR!Mw&48};32v^t26!emm^_vW@Mx}5B9NTX%Ga2hv)-|lF$yhB%f
z>*mOa(>)kB<tJRWyD`<-9NofPIl3Z)A@QyGwbK&Vcx6*zcM<=H_p0RCB<9wpalGMJ
zzW%Mlqx$~*`|yLXyKLEXQzh0~?8Uz9I#{0x#jD@8h?qK#ZS1CEd)r3LS=Wv4vf^QW
zw+%<hT~v9z0xwSrqP53p=C!Va`UXWV9VBd-=QB~iHjM+WcII)%XL#FJ+^wfl`NmnC
zAY}^dS#HNKO&s{Ti5`38*CG7AJGYJ0hVsByU|<9Mqc7pD^>F(Av__uHpw^_D@U2z>
z-3OWSrLh)+x*vqrpLxhSn#6g3^s%s4S2}LbU`?&~x07dJYeojnIlM&In%USsd@|0z
z-+`ci4Oki6ie>F$Is04)Q<nr|$zgeZWOifqlu2}5-H)Mhjto*xW!pRQU3M<uh;^O#
zaefNU2n*rdW*<y4%VMxNiLX07g~Rh=-0v*29Pyq0{Sty#hXwhvcqKl*?a4Wf?{Q+x
zc-a-VU{IS_@vu(D#rGF*a;Q73Hh7^Rx+5A|k6~@D5pK@i4CR_adKpioL-K2UjMalv
zkT%9oS%%!$e)KwTFTPDPo|Ift@O)!b=?C%qkT*~p(2ucK<a?|!1<n^W__e$(-Mi&u
z!19Nf*!KzYg&AjYDuoj&UtxN3A)3!oW!U0>XuhWqud)V8$1D~4?VE9RKX)!TX2H=i
zH*|LQ;JInZP<+1zwyyG>U$Y4AlI?x_!;kL<-^GqaN_<?k7R_dq<I1~DFz#s0hWa8J
zzTJSHnOgig=@raPyRoN*F15BS0BsBS&#xD>9*F0)r~}>`O=6tq7-V?P!CmFqsO&ri
zOBV&u?dl_ZZ==hMKreVU8nRBK8~*Nn1%vofDkWHAe(XpFY)=tCrbIUTuE2o4<LDJu
z%wOhv(7#`A&N?|)IKQdLzOfcjtMt&&cMP97O~gYzSGt}5fQ66Rv0#WRlvfABea}IB
zbQ{fkI?+sNJdGuj&!hJ9CG0l~rurO177jOIfp{*SrH;hQF!1l1ZhWja1!G2C#QDp@
zbv4ms#5ffk@Ge2o6zQHF2*K&xW^8!a8b4$oSFxr!8j-+lmL+hG>c)8y|Ka|bcHFUD
zADy@Av9sje-8VnNMh_({Esy4G*(KKmO-1*y!hxR~&M@=O$Z;8qezQv0)-Xc+3)y^r
zzy>>8dU1YUIZ|?ExAaw?pZqfMtW9sutayvi?30K;7R4sjxp=eaAdc>F;n8imurBOQ
zgMI28d2b?fEPQy>%8Eaq>I#=IovEYFV|CtMWF8AgNy`XsEc1cKu27VV+ur?A0;U&E
z!L~?WmL+=AY<4rge4WT|4<hIlVb39J#?Z2rIAqPTnY-VM8o`4Y_2UtmXY|0oDnr(;
zGl2DzBa&5##Q8gIaKlZWMOW3ZwT&Chn`_{pI0-6chnS}COPAvg{I7%b)(yI`vFj29
zUB8V7lHpy^wGUimUo-ogBUYbCqT)E=U%m*yUiY(bs1nB7g1hMZqXYlW@nYh1BMxod
z4093_@vOfVpT1lUhlQ6gOm@0=`e-5|xGhKg_Y(UCSn$dq4`}z9fyA)Ed?cCd>+hq*
zE9r&O*OORcXOF)Mh5Xy?5aP#7U`}nOyk{iy+2tR;IHmD)a3Qi=Rzm$vJx(5K#kF@7
zxGdBH)$KOoaK|TjwW<y+iZ0<}oF-R}%tZC+d}a++;G<-Jx@GEN?{aUZ)q1ey;Y0=+
zWa7?Xdv5C<&!bc4W7_;o7R>hKq2a@L=$#j?tkU7Z$C}LBYXX0#Irz6}F?{4K<Fvnk
zg{tyzS`*8L4i2nnwBSm6E2e5XV*bzw8vYlJh-dTQ+C_=Z)*BG#Y|C@{G3cY1!Sen?
zxMqhZcQmN*w|Xa@PE3J{X+CCEkKmmE6Jf<EqrAsuY#Mb9Up}X?N~am^&J-i@%^d7F
z(@*A#Ptd*X0fZdt%IPC3Fs@%mhF5xG*bjG(i2=~%voH?&;l<=d^8W0NCNm>ovBnKQ
zPL*)Ctp<Ny>CC6~r5x<sjhC+t!L@}8;XJ4k-z<mIZR8l1iHEXuWNUV}E#dgq+4O2T
zmd|DWWph@M8BOw7KhzJk9aQN%C{VhhjyTZs2lV3iKyRM(VONdfoc4uu3~0th(|{d2
zBvUb^BkYol(Ef;IExx4kjzTLmd{}}o?J~rD_2MBt4Zf4=*sO`e+31iVnJfp%?7YUL
z(=k-tnS))s61cScEX;h}iKgfE@K^UBI&~;PQgjsp3Ttpjd9XZRTfk@XJ8TSDihcJT
z*;mdM-PL!YX?I1MZR*D6Bi-2id2jlUQp8<_$Efd~hWL$}5tN|E!(rmjSlknj_9epi
z)LI;$zYGtq97pHv=W!^mh^BXgIDBp{ze&z*+ssL<JK{&xj8sNUb3#Fj(X6^N8#^R(
zvuJ5BKKIoXMo%Fx_pe2Em^pmbjAZAWG?w3x`H0M`eCto+%?$<azoLs}#-^y<7X@uk
zbIv>6p55ARMsTMz+EnNW|3HH)bcA6qck+n!U+}zM?k0{`5n(=-u9=c?TJs9C4t++1
zQ4f~i3}Zx0dmL~74XWY`EB`IokN-?~HOdTz8(Q((pLp*5PnRvKHKFxyq+|=+aI3Sh
zrN^k?d&^6B@8BWN+~K^J)17_y`Eb14{Y!2P;<7%fm~V9+OO&)Rd#NAvk^}kl;B%RY
zoyY$F!l*+7u2_&tyDoKjzo(z@M86_((g^GdOQQE74dEJn#LEfu5tcX$K`j^J>CMqR
z5gCSN!`I-N>TDb~v&J*;%_!7!#g@IY6JD*#DbL75v+f9k<GkFH?dj!Y!-~20;reJC
zpZ^?1Ys*qz63(Fcoi?mCv_)Rb0H92huZsKNwONWh3%W7zSt4g14TIS>1=h4449)S5
z@GuZIYgap7yyA`}!kru8HW?)vanv5G$Vc^EQNDc=^y6LeL^Bn$(}i>VZwNDQX`y6e
z4Sc*Tne{w?O?n0JVAEo3J6(+7<HcE5G)cTVYWV6{kJhgpao%=-?4>Q(HD(w$wDX~1
zKSgX*GUb@gU(x^85Pnu!EUYL4s_z%(?$CB@y-A#SZgx0rvK{FedtsS!3TMRKr*&Wy
zs#jREvZNVK-1X(28G{+o+?DmyJfY_G2pMwDGR)cq^PE&>4qA$;#No(D6CP=DEarZ5
z<k7=67&T=M?uGfFP`e9e4vONtnc?)3F2(JHTGYhaqm9)F)`b<(!}}|CggEh7tTXOt
zb_7fEP!np$GuMkaWHva$Asa*LuVDd#p<h#v;cL}c)WV6E<_yMptyo@{_tHa|2RnNv
zu(om(e%1sa|KD+Jy5`MSKEr9Z_ZmK*UxH>AA7ky*Gf4Vdg+V{`xaYewCcf#$nvLSy
z6z*ZfE*CfnLn*o;4{zTo^Fx9S#@1gz#C}EOJj_5+E6ELBZ;PDAEwFY>AFN(-5n4`B
z^qn0+Rj(^}CmpHS=8}(*{L#7djwlE|k3Efl(aBedea}tB%(`2`oOh!7k-zx9#ha@a
zsBnt>yGGu<fG;IO@bArYcuBV@L17p_b$7$bVZ}65pTtvlCsJ`$Tb@wahlqKf(Yxyx
z)S529oT?@G9k&|6&f454b4<?%xv(<oh;wn-EWPDLtqyy!`ezl2y`;xH@Fc#~ZpDC0
z`e?Fl26k>5$485@cp?^T^z00iH(j}Qi9P4mTH~bg6ntK<M5~4{j8+)O7pHxAP3Bcg
z$5bMDQYt(DTZcs_OgKi^9(t0`)Hv!6ho?^b?l6Hz56S*$t_hFnS#xNi9`26pi_5xN
zY&>RwXWLrvXSifQf9PO&e?v@eszQ$iHu&Dt3E@hBOK%%o|08FO>7i5*2GoZ9dH9y=
zM>$UN(4lNDE(pVq@xtu8?<v`(b0~CQfVB(bXg@p!xoe8C;o5k_uV07<huYEJX%nKK
zCsJ$7beK(4WQ$u)xLM@K!_ou(V73(VqMNeJ)fE~ln!<cnq(Y+|YvzP-x}o^c8YM@V
zRX|mRh2o&tjpsWY(2(X0<--x2v&oBZZbis0eF2;oioZrvm{gK?e=HtMlXLNG{MQFY
zD<u1)ku04VGiL2qWKQ2cyfHEk#xmav3mMA!gRD?-DV-V~{+O)#15fPDX=0QpJl@~f
z71Eb`YQor3@-nk`-@<k0G?r%d;;K#Vth?6=$IVorTQ`7fg_oSLHiNswdvtYvB#rZ(
z;O;MTf(dW%euyD%Uu(rLPCnEWo<e=JGj7%faLv%ESoO+SI$kd^r_vXztZmsCAv1;Y
zNc`1o&*`tP;eOaTjQbviprUwOe!Bx&!*?KEuOpUpcNDI86s$sSp{e#*naK~skMiA;
zSu;TJ;&8ZGErHY2c05}Xi0Q|F;`h7)YRs60*CwjGc|e`MS&j^o9nH)nC;pWl6J1R)
zaqJ`H<do1|DV8g3T=2bYCi)x}UR-By#&%nQ59aGIXU}kYe%J%MC%$-Uzd;;kaR@h!
zlXHa!t_;yfi0r1!`vmcNxSDh$TOl#77oUI0VvlXD;5w`V{a*@iBY6-`ZBb?X!d1A}
zTl(7#o3XiFD-2N8#(*JTvGI-{iqzWi?+iz{bo!0@5!HAt+v_U_doZd+30CGwhi39_
zT;IM8mwXh_I>Zw*LiQjsp$)#J&%<-^m)*Lq&qc*EQB}PX$;QIJFgS+%Bpn3$h*M8k
z=ZjRuCv!QS#!G);`dDM6)`@?)<_dz+-@+<z4x$}B5in*v{2O9%qu>-CYfs`LWnsH7
zc#Y#z;uv)-3Y`)saOL#(uvxqerE=H(@ZOi|b_1|+ygI8#+wp{Ce5cGS<kdaGYdn|7
zco%CvG&qgJgU9pzw(nTkPzr^AO{CL*2cKIM;zigl*pyj7wSOW$SBF9GwGYdGKftHC
z#atQx9b3XP@ovo>^m`Z0>L0rFP1l9LWIYQfNjFMO2ghbF#I1Q@;*cbbE!5~KoX4Qk
zo}9bL12)6+5PEVJ27Bc5QCKw0PA2edUJ%cJHe$^Q*%Jp>LHD*R?|7^8qs-l1iqm1&
zMOe2J)*|!m0w{iy`L|yz_t$k`i}fq<)+CUQj?yPL&Z33>DL7dqLHYDS6h-RNd|fXt
z?Wri|RX1Kfp@qL$az^zYkHDB#G+UvLxox84XKup7Ghd-xcJce-)!9boqE3@X$qvnh
z@w?w4yJI7sv~ogjc?kCiTlhnD3x0oMkG-X@aU>*>ueuE2?bai?$Jh^*<%78JVm^=W
zbwTID;%N|m+Ej<57_hquo9ey5fR?(L=IF&Wrp@^%s}GmYvf%3RBl%|8G5B?HU>BuG
zUR;rd(-yLGxYdLCJJgXmP#8mYk?dbw$oe#K$A3JJ=E89NG%%AMJ^OH9j5_*l&ZU!j
zYxr)h!HHiPXc8*9hmQsf-fhk|tu635T@&NPt#$oA@py1FzgaZE@4Yw@zINu3EHfHE
z8-N+hy7I`ESf1*pfCCr1Ks&EH!!ymf<(VPf;w8`A%LL#0pT)|UbJ#Fm4Vs!p*n0Xf
zqW8E%Mfn-N%n<kHnX9O6oPiOo6WOtwA#dHT#!+!<Lf8qT3-)1kyA0SaU4`NAOW5+~
zMR;WepnjPO+nQvvOwLaO;}lr3NsB`i<o)<On~&apK!?qm92c?{XM0<t@JTx=``Vz%
z_c=KA)P?unmEibBUFn0RU_*>_Jyzeql6&o#(ja|>1GQ*oW-9*4xu`1F!akK89()nS
zx%0ia&(<FvWfAmUri;jo=}5`4<lL<@5Lwe7u~ql*XZBCnd^JE!z2uS8i}B==aA3-u
zG2&V|=A}vQIv@o#H#;K2U?4Uheun{ib<(LG$QB3Fm@Ap5@IN|qmhbetF*bY>yaux}
zOrWS;FME-mP^i6!)&C6=CP$d;zedtWrvt1<E6B5VG-k?7{Y7_2c9dM?O3ClexZVWY
zo;Ss<>&M}GAOf=+{n*%c5|39k;%bLb3|?WwJhO#xdhrd9rdMFPXFDu?E>2@DQ?6Q)
zN`=sDhKD?Zk@<2owQxhnO?oJ|>C7Sf^Xd6D1B#2ya4LB|{x*r`y%*`wwfEuVrhBkw
z=@XoiXP_U2VY5{o9i+eJnxGGD^&K#u=SkZpsTfH!wqG8}RW%9x`|lJS7nm~Ca4qhw
z?aJ>D<LU5PK9?5B(C9LPYe$!$`AlW>wv9up*^;HWD;}IFfv{Lqi64urF>Fo>!;TCT
zSGWc*%Z|9;<v^Ir-SqFM_S~{AAGW!(@#eg;yxWIhLFZ6r1-8b^jp2A+F5U}y&+JRJ
z=gssGR(xrOqwn*uGbEk?J6a;;wIa$)#j9%h3G3FT(k(CwaRKgFkr{wZF2S54+1f^p
ziOe<Ig>xYe?A;KK#_FSZT5}ACMs~!D<705QPO@&lW}?%Zg)qE4jBl^r#_C1BP`p=z
z8P;C75G!|^EtzaAc>>O9jbG>A<L(JzT3Nr7URNNUe#vg9y>PX@=(1;lIq%vJq4Cue
zKB!fqn(XxN<*tLJpAtVkoPj9k8kl9>M_c<IG+u1O%d!`7E^$Ni=?-x8cA;g5mgLbB
zSp07Wn#5Ma%0`KsrtQWDEQ9(u$uh4P&anzzF|3UmXD1KipLgowNt3z8-|N`7%Nkja
z3_0hu8V%cuS0}0-yF;xcx1GW>K_zG^{0n{iwjA<bJB&T!Da<k-^m@Jt;l`=*K9_I8
z0|nG64#76JE2xTUhw7KHI3WEsi&b0E;g2$(^t_0i4gIn5XdX6naO0s#cd*XG2o-7H
zp#CQl{hDmY`5i&{JjW2a;;CKVHeY(T6S%o!cZM80k2#*w!#;Tcf%>U9Tzyyexg~V+
z*Q2e>x2-)J(as@>`s0!$&s~ew(gmN?t_D%(LgC$0a^0c(;OXbjH*NGVrkw?qM<*lm
zb_x<ZnX$kAGbB`g#q&1m7=3j$E^q9@;V#-Jo?b{Nz4kQydIYzd58xB!SVUOnGT^Ut
zbT__4f{6oi&S~<7i2~o2OU7_QdsJqIBTris!~7ets6rQmlTEPcrxnkN58L8f03S)m
z@$8Dd!ltpMnUS!f<_Om(#ZGukj_56Cu}<slxN7PYluU`Bk+8(8hZ^wWqv2fc(~7xA
zB}<UJ4YMCb;Nx#?@ieu<TWo-5Wj{3Q+#fd+EpY9MANQTn#igOetdZVm!Bh{#`yasG
zcEZ|N`Wo%*#)zBk0OIy^Vyo(jm|ocq2`AFI;p$L6%(#w+$GqXOe+DK?j@FK+k=@gl
zUZ2MBkLEViuO5IFEwX7U|2_WxI67}yj+?7BvEH^bHr%Yow(gl|ayb?;*YlYDw+BoV
zjj_yP5uRNt{C_{WcHAQj+S-)MG7q4=drw}KUT(3fiSVA4QF-erte=g<soQ&SyL1!-
zUixrB_F*JUih+Uj;?vXo*fww~>{hmCl0si@S<{=&HnVYU<2$TAW(u{HW6)gzoaE3K
zrfnMVY_Fnlt7kxSS%9#bE~C5uF;qq>qK*GzVJO9OOT7_OPc{)=(P$(;OJ#4F9p{`7
z@13}so7uKz<-#HqYPi8IVgO#$`ty>F4h<{%pgLzQDyk&M)%+SFg^5reoW{Cc58+xJ
zib1E7cvV$_3qE_|vSk=L$WHU(L~$#(=qx$I(a7E$%noWcywJ;o-?MwsGI{}W+B#wP
z32~!H4s%JZ_y(go@v-obB8ImSW~wVJYevFjYB$&|GU4Xd>3H)ii1|JA@oJkM_dZW%
zK-(T{5~Ypa(JJuGYk>MdU3?s+D&C|4sA#Xi(%qxkd|Dmu&wY#@&pV=<f;i@N+^KV-
zI}R8MgWGKoCh66~-|#(FGJ=<M#=v>2CSOJv@K~+_F4j7T$3PR0-_^n6`fsHBEx{~d
z)bV{JCVbUo>ah`s=%>VSO%u@l=MbJ<-b%b?vJVJt$;yp6xGnP`9}7QW(Fe0=l^d_s
zp25|r7jdm>Gr|i7h!dzkwJxoO_w7I|c&~-&TY}&<Tb*_qN%(6XM{DVQKGg5RIS-Cv
z=5Bv<85_p^ch4a9l?FXe{K4qMG5ioy29IFri`n+T66t>h&l$@_Em|>9s}mxQd7;mz
zLDbLp#>RQsZ2iQWC#7?te8iXQtJXumu8<XTba3UmC2h_bVY&1zpE&g6P@^yu=(U9I
z>zU{}w=+)&KWdY`8eSfCXUj>)QF!+dR{N!}uXRT#_fy4#sZII!ptWSIW=RJ|jV)?Y
zS+Pa-i+5ZY{LK>O{l3AeT5<sLe5&Y@!JYEiCFy=dupfEwKzFn^@j$^h4IEJ}Mz^AH
zzMSaD6_Q7|bYeL&HS~G^TWbz-Y>q%RO^lJfuJZ11;hq<wz0NNr7dGJe%0f1l8^Eh1
z3kPpb$L#tJoV#%gz6N#Xrh~$iIMS7s`~2wDR1bbWYY~`lg~^S+v>VX_wPl)EAZ*`+
z6~Wjd&mARy$zjz+p{}t#HhdhwC0#8gw<x~g%;EIvY6AYN!9S~1$qsbIr-=3#<#}B4
zj7l6la~MZ0b---BM7}8M#*MFg%kQZ)Srm(5ER5X+p?@(_I#j=;e=_#C6_XbR37`2N
zO1p~ZRB1FH6wkuMCE3*S-h`mvGjUCLBXN}n(Dz&t(*~OJNnlqj8EGw<Ml0Eu*`xAf
zXS&)rV`g)4h1s@2Wx5JJ9UY9<Ia{#tkvIbs2C(ytD(H+H1oLyj^yxPZHxA2OPPq_P
z5mi_)%m!=w6yufKb=0i2;J3FatZ1nsOza=1l9^3%|8!p6XOBr&6nS%|0~f?Bz&~Yg
z>`GjU;LjHL?HI#Wc^0@bWego6rE@H-t|oGy(`j1F;-zC*w@e=aqs>`W5H5_83fLNm
z4|=r?CjQETLH$Ukg*9hkzBwnRh10zI0oZ4a=QgDu@Vs&d8(ZjLwYPL|tM+2$RCS!a
z+zy|MhSO+jA~l!qgsyQU##Tfyv#Tp!Hcdg!&!&7VPHzu$H_jja5VNw9d2rftbibpA
zp5F$ui@EG<v!kiHVIWg1x*>ItKf8aNjt%-vS&|Th|0?xie!_w|s+}3WZv)ooIpL~h
zGZfbEM%Fif_8U19r|&A@#VX<WpD<%VhC8xztTD4=5Bh$+jjR#f@h<!)4AMWNT-zTG
zlTV}DK4m&9bilx^l9d$a=d`g-tlcCtq{=RE+7ZOd3AQXS7w*%u8Z0Z6o$mwj4DXg4
z*baBvOkE9^p06<RNN)r^sK!FQw%D!Ff%;#KB$H%|(w085qx=kQ`T6S$AH>u*k18rH
zF|o`5y+hk!@m=v3v>MLDJ?dOl8pd+}t9bUg9qsx!<8c3GbQRC*o)tS_<r~N6-970y
zvL`oLb;CQ2?!q7Fj@ZP}oI6E@IaBT+L~|0Ze@#c~Qg<ple}JZRqwOS*w#LPlcb2rm
zp?16R*`WzNO?$BWSTk|<>GD?x4fGS<M2_BG9GTXgibFzh%-)^nQlv|i5XL9l)#;Ho
z6^f=t@cfX3?#;A?tL%bL@3gq3qZ3!kTt0kp6Rf;njP~|dVfZQo%SN8XPJb)t7q#Z$
z#&@vt7S~&dtIP>+V_m5aWBVG=JhvK;KDFfb^0Cl5DYLZHrPyfu2N#qKc<8%5LnYU+
zV{#t6gqhXECsaB^I#|5Wg;RGMGg-$CTPDY1?e6}}&y${abO1*lo{llmZn*HH56hlA
ziN9ktnx9ug<k*+k+e_}H{w~b(>A@7GOE`Zbp4*H&fk7%rtZa=2Z#Bft3S#naL*D3Q
zLx)!V`FV7EmN}}^D%pkqp6X+A(Q)j2l*~K+SqL<54U^T)5&f;XbXngaw8R;2?{?+*
z3?&%cx{u8M(X<>J$0~Wg^bCK7xntrO>EH+7srnr3s4qL&FUTHfg)8IYIqcR%)GFGb
zWUMb+$u8*YhKCq_Yb$)VXkw1}7o0S0!N@Ht?6%E?p(}PGD_aGTT^3-tc##w96KN3L
z7C$a{bBS(CaZsd+XDJxHbu{>3*Eq&0IpR)$4-Jl2!_cq;XIc92<%vjER2N{TMIxtm
ziKLZ7B=(J%$dzr2P&?ZgyW7;^XSD{rLZ)KJsCGQrt|#@^xp4118(ce+go#~j;In!+
zE{@j3*DD%$D*T1=Rr=g}&YlsI%o#APE9alI7apMsHx)QCy4aPTeNyPLR1?44V(A+y
z>|Dun*4mmP!e#`5g0wMnMG~i2%!T$r>8#C~fez~Fyr$F=LnkSq>9TaT%L&KUeOdg{
zR9vGb^6W~u0Kd)65O}~8H|uwz?-E6v7T$^-cH;R%12j*I<&`*3bchfJ(`$8nY4Tp2
zb4^h=Rsl~>`f|Uyr|d!6VEV0dxIFf+SCfqwP^empS+aW_t6|9h@<(#7N-dlw58#_3
zVbGq<mOUA9L#GEn%Gq^DZa+4?zXI<L_|bPr77JPqg<)%O!`3$_y_U!jjsDclHpSgP
zXE3w&0#5vO!uOz-82lj_i?;T_-2`!CnW-`MStz$P*@^jcHbJjxOE&9QhTeYSX&RPB
z<4A4RjuHRSuy8Iu>c^OG?fGoGGE8lTVdT#|{%IO7t`l`6+E(JG<0#CM{_DF-3Y-)%
z4MQi(pBJ+VV`5su>ceFeHuu8TaVESG`wz?IE*YcU0oP>*UN~F6zwK1n<)|5T|EY7y
zXjAbOTO(Pa8MfUpV%d%MXl7%;cPS>YFYLo$^Atwyj%C@8@mQ^z#xm*9S?|_Hi^m@5
zmiQhU{X99$-VT-Pn_=|wIdHU6MCT4YsQqKMxMtH)`7oN-zQ{ZMZ#UYQ|B+orG}R)c
z-&$=Aqw&2NzHS`-W(Lu=IGQot%VGDr0VAFS;A*lJ)JIlg!lw~@wxT!YG&N=EoVIZ8
z;Eqq3a&D;;o^d-r+N8_h>0$>g+gOWv@{XB0%!daSy1=e<JHl@LM8W1?&|j0s&R<Qa
z{K5t1-^XIqrw1^z*Fw*;DLf}L`@-ghSeO^d3t!dITkc>-`c1?!zd+6$eFcY>DdOXB
z9qRs0;ftiUm>B24BCmRsuRVz-6BOV!!3@7_Tk$lt8SZ3FZTZZ1%1kNVus<J->dJIu
zRm^yng(+#;IAPig;V*h&-sZ>nxup-ky8AO$r5QfRPV{h3cNlsaGdswRZ_B#ztLXul
zuRD(+0oic8tO3t06JT~nRXm-!Y_DL;J-w|k%l;{Pw;aJYK|2xp>Hw~NQs>vGmR$cc
ziffc~7!dIWJBO(<TJj+cD`WU?t}yC*x^wS<Q@C~LCqf71LHT_!I+@PK?_0ak&~FOH
zt+3;2$?lcCu|aK79$(Fwgz)v}@HJV37ScKOkL!$=x2M6%$qYDAh-3eBSzh2nvyz_}
zfA9omNDn!<I-29BsUvKU-1oea89P0gxh*!~^^_L4kQB@`zXVpBX>pm^Y((uh#;0{%
zsCLVqK6kA!@6r`K4Y%QfrByiB+?3VDzYw><m@R{{G0$rVT~<jaVNC|Nddm#T#GkI;
zJXoDk1`hOrNn#iJZ%~FyWGODCeL~C_UCeQ5#{-+%LwV+3FRvYne6YI?zXz^G;Uj-8
z*ApLKpuDduZe#2E5WYW0?mL-|yzPbjbTx#@lB*aOxef!iiMMO~IQG(>z<S%$c&-$~
zZ=+Vi;oMam9u*{=zrPq_a2;_aMac0`!`=3dtm-=vA=lL4B|Q0+{?6DkPiCCLs9K&a
zy{yx+r<R?b-LP1GI-bwHONOINc%CbaPvA%CE@+Rlz?~08c)v*Y7#6C0|2~*EYOhJY
zG?r8LhH-gjDqE?G?@>4ki-hf<WOV@-4-Dr*C2?0Z*z$p#caPrm;fu>|P@B66TV6ZE
zW$PB{B$`mcZwSx7b3&Vg-*JDAzxcS`Veqakc;8cz->-X7v-tlSI}@lH+kfvj&yz-_
zNh!_q-2VQgNeO8{gC>ef(j=6zWGM5Ld5%np%tDVLM5ZT%RHjPCCy`K{pL5Q8*8lmx
z&sp!g*V<Zp>Av=E5BI*W`}%%AA9mB+@#STVFwzXTjOI(~SQ8pZJB@~8zP2Os*8M`?
zT4S=Bl8jYPy~)(Ti{8kXLP$mdbu;zwBH9~uM?B#)Mh><8Ij@kXi0ukG?EHx!A)V70
zuuhyBv*dB}@<FVtu_1%AcVIZC1O4A`g;nT4yfAd6GihfbFlE-nqCX+;cnvbG0-9d1
z4PO16sqsY;gq1vzF<uR`C*)DLbvX7POhfvGA&`Hqi6_UI$(ghZ!6Ht)Zl?&3%)y9w
zbOtsEg~HrH6qf8o&!>dbqp<>tj=lrAI14h^v<3I)%A<Y3UUc+HlZnr1ILjI$`7^ux
zuIr&C%ALJNGVq&jOG}PNl5v+P<hs4#^gs_T<7?o#(*fRwS5aphPAfkhh3cOoIN|Sv
zl~smN5-&$yuOw<hPUEuJ5R!Z6gP%1fbo7t{-OZ1oV`cU<ZQ%;Ymsldc&j@MflA+K%
zi0a)$aeHe9MjLA)Oxy~aX1{>JE?JzY(&b|qc29H`qDgWbU0p9gQv52c;eO+1c706B
zGT{F1ek8r+e6`&Qm=t*6qY3jEcje-Es({wX&Y=mua@1C(PA|XZW9gT0%CoV?eC}S~
zy~hsw@e%ZSkv^hCWS9k8gRn6}NFw+?nqA^3cskMHBMYGtX#^U#3{Nh8!5)_(wED;l
znsRsmoh#JAhxrq+lX({x1`H?h&Xd>>%?yYFUi(ULBsqmj)cjh5A6>Fkcw;J+rH+J$
znJn6@BOtc73#I!<pr^%vuEwfR)GG;!*s~5f6KrVpesRiox{iJaI&fvL1ubIkZBC*k
zcJpVToX^o@`tY%}wGLLK#bCqTO4QoCfQWFPVEvMrIJ-CzIa8-_2Vfn7`*dmc*J952
zOVE#}ZD`g@q<g#Ab6vyFX|^+D*@xVGX9$f+KhKWm!O+lo1&gx=biyTuew>v;ww5*q
zIeo+|O*yP-mqWGWP*@h!V|=$KjH14vpS?DD9A{><i6M=0ScbCO`N&OGrm(x4P~@db
zQZ*iknm3W+*d?QMz#DGXV=y5n4=M|k=+pKTDtOF}SW{;*HP?dOAxZeDnUHJFGJG3&
z9xuC7`Mpqv>kmRvsoaH`ob7IytctqT2~@l-o9s-ba7=wNI)aSpNwqq?x7VU#XQGqO
z0%$zD#iAtSk?!^gMd_+^^;H>k{PZcqPm!LVm7+wM41CSegx4ekdT~1rQ*46~Y<iP@
z<w>|bMwaeYSmODQ({TTT+1hJ+1xsoT>BNR@h{-Ug)W`|w?|1`w>~fjjWJgX`)9BLR
zjnJORF1bKYx~q8?s^g~8{b$kS%Y7G9WyW2~7GmMEBINdMg;KE#WfvSpI6F~1YlPS-
zH;*2y=XD=GPG6+qgRw2cG4btm8eOFaCl49SQ1rnT^)R}-z>|I&^Y?cCG&-_WkDjR5
z(W!Q3sy$eU7v8P7ca?cSqd4ndxe$0RK;yGHuyNz>hP=H{AF>PglKj|#%q-x6?E9Q;
zjUGOh+q9_y_8wZ~lr$ZVKSUr?dj@H_F4VeA2g@8(uy`8tp_z*tnxsuiK4THb?+Fc_
z*PIY>M`zD+3}ug3^^RZYzqSyW+m*1t_AI{7K7gYihEv^>K0*IuMKGDHPp4jwBqe)G
zgpZA&h9!^TU6n+mYTaowdo;#|NRed#^Ze)#6l<A*HPf1U#~4$bDzCFM+w5oEAlQfY
zrzE9wh@O~7i}LI-J@+u?dd87=?F16$we!&V%+z%YqHF8VpfO}MZXeTQ_ohExh@3(_
zo5#|rgJaokqk>$z0-@bK*xQl_m8mOWv`3!g`clb{y?woz;n)>Ah;oOm!JcccaP8?z
zEa862IpMi9{Z%A9Lg%48ke%;aN6@fxT|^i8pt`>ZJ&gMcp~V{1|8y638ouGmg5${B
z_XN#jBFX)X0Eu&%xji$OR%#2;(Pr-OeDI-oTMKGP{e}q|st8rr#f1L)h~9n}ogW0$
z8C-$B79r?Q<gWOD6k6$Z9&yt~()<Ntkf;_zZ%?m4oOw^*FT`QrLw_WejX?bFX1ofE
zz}9pr?Agbx_l6$9G4(*aWuC3VjKhdY_oYE@N04vtPE%@SnG2ywB8dZV+wmy$+t_9E
zfLYsop09EJ3OM$3LhBTJWAAuy_JUbF!EIReka^KFd}#Q`IOr)D(mw8TiN-RsqFD<W
z_xEDL2@4u%bQ6)LTj1*A1B*isAji4ph_#$wsT_$XAGK*tBzJY1<WN6Tf|+SG=vr|Z
zZCi{nkS0*tcmXCcyL3f}DfWh0(u}DZw3<6M{^P_TT~!5>Yh%bcQ4&h~##7acU3l`X
z5uXE=A^VsU<;n-rRGTq)(-%pv6Z|oQGlpX|4j@Bq0e4l0k&M}3ir@SN`*#M@{MsHE
zg?Q2a7^0N0HslwYGo$4j_Wuz^l?9d*8?qZ85|Yu>zZ^xn&X`hOfz6qs2)Z){MnUcP
z5Of4ZCq*$!?=JEe*|OV(S*@!Zap9FV%2q04X0jCGcCZIGC6IoanBYzHGpJ9K<9Wz<
zY6#zii>l^m%2c35?SC>0R}sN;7DGH>8?&V|anyDpI@pbU_njK@vz)0QDV&b<bHsyO
z?)|@Wq2OkHvQQDG+t!@JJYz>joCS(oWkU-Tl_~DD32h4<MR61C@l^39uH4n8yZs7q
zs?-#5b#(~d?2OO?X*xaW5N;-w!K?Td_wI(#y0125^+p=i8BGxOt%P<L=LGT+NOE!p
z9X`(U$ESUGX48V6W44qt$`H!O=fOrpjq{?mRJ460=@h3jw_`M|BSU5dN#bMvOr(sm
zppW@3WVW>n$y1%sz)btPsQtKgXCDkNJ%Ug9Qmhbmg<D!X{<<WBzlsBBaDxM_c_~U2
zo~N<+QwGgv&UsX<3@lDJqqIeuR@4~M`rGl8b4?NNdse}w(vtREID+}Pa+ot{4jq`(
zkM8G&(vJ<!G%{=<BH0rxwK$b@S94yA*_mJUIFJ8Xg`HU{c&!vo?jkv)EaO7CKR01%
zl`I}SF2d{ciue^X04{T7$(wuRt?ZTEWMfD1{lvM)egh3H2T-pq4e92=^s#}@14Zz$
z$rMAfP<W1;WfIu4&VUX?{eiGsZ{YsJi}L&FqxIW*2*96pS0!}rxrq_#mgM(Sf?cN~
zcv~+;cUEt~@ZzQTmS#jT7sP3z)>+O{jK)}9LRS7zJpI6(wTHv#K6mBcX_(XOS<W<K
zs5Aw|$WwvX3Ixy4qp&0~?ztMEzcJ5IOTVN1%|voGlBUEt=GYb<PN_lO)Kp*w;~|4_
zc$yxzHb^4qK^H<MWzmuD@ie!{f&R>u!Tvk<@v7DbN&`!v#QUeUWBc=3_Z)gQ+yv1H
zb`&#%8MH?f$?Bpc#*}DMrw;dNe3|p2p@dIC*>t@zl$JBkPTq0`9g;4@+57R-@Y4%g
zVH)_XMRb83sfrN?kfv3InTlg^b|Z6rZRTLAgD(vnEe4g$DYQ&_B&Isv$L^6bcseW?
z!#B>SO!g@ESuI8_cYQ6ZMQLTIH+Ah8N|VHA;)aw0$}E_1RPh{dR>_mE*?Mdkm`=ZB
zgxPyjkHfLw(dK>`D+cbwWSuSS(JMe&pC4^m*o*fU9LOVoAaea|X_JmF0*eHs+R-cM
zyi|)vZ35CgDFK_PUO}ZP`*het@M(@MRKIJGgh(QHBRI>xCll%l3$R4Vo+@sH(X0F-
zJS%R*6L!QrejH2-Buvn2EQL&MB}|+(ip17LQ_rD8{J9%PCtLCnG?u$kE`#yuc0V#!
zoe$Y(<0x;}Q0#9qA?xCyv`)^2hK+NhAaIxCY$E-sU`$Sb&45;2AoT|G+)I5DwfomY
z*+YzCguTdavlP@qN79)~gYY)~G(LTKiR+nC2$U$s&#l(T+Pn}o8=Wa-r6(QGu_XB$
zcQN#1935CR7~2cHX^%@brSwWs+L{q~p5jD>*LV$~!Va(Y1W}zq1XXW!q^u2FP?bN0
zj7){FF+K>7_<Q5r2nA|;T8fW1tRRu7N7-)S_@y}#pSOKQL{S-%41CCzMxi5j0*N|Z
z##f1t2zVPs4HL)E(ZIE+Unvi>8KKx0?n%+t#hKkbfzDGL4uqLgt91}r6wN`W|6ufr
zfXp_>axR)NWNGu!a6^mpf6`ccSb-9iSKzgw1VYD6r=BNKFpiKzpHdsN)}-O6Q6^Nc
zjmGTP$I&o!3UxGAU{~#HcsVbJL6rcN4;`uAvlVB)d0@iY-KZR`P9GL{Ks7`YCV^pi
zz#PVf-Rp63>qL@O;GAr$EjvtSkbhVS+{NXoT}1<bl<q+nZG>omD|Z8|sDZynE<Du6
z`tRyAi~05AGk85oH<aXCp29?k-J}BrNR{Y--=Gr+wq3&exoRjs`-vxmw8?zECLdq*
zry*t8Bsi>(AjcT8>U@Q1?E33wuYlxz?pEfS(5|<6Sh^d~J;ZGKYEPW|vJGN2#+2N`
z-sLjhZ@DB>O7|PoU3MVvlb}S^X|!|YT4-J00E2V+khji<H}ifwDojZ2_8clMZo>Ov
z4lur)O}$6WApfcXgXT>lgZIB6{VATey9~`5dwSWJMIO$4UOAOBSjEh7-t7Ua@DLiZ
zLJC6ZJnvmJ0k=jMqJB;{wmtO0*O%KMsV2w$USaN_EP=*b?oP80cet_xhE~g<tHFcr
z^<2P5W;V}h&7ftAOE6Ms5Khb=K$Vk@p{SU1XWLI;!F*A~C!K~|l^W`tLuntgtG}fr
zFyDe%tsNKOvzZwOvIpSyeIi}S23gJPMUi<J^8q!n;D!>-%L}H?sD4NuZ;hSLmtmOo
zSxEa|hxDt9=o|167E<iryKYYliv8$PXs;kAR~5tlOr#Z>C(z+M3MG|;Q2F*CBJPUN
z(Th`P_`(z<yRui~;vh`4aiL??^N=H@g7fRRvw5Wzg*(+~D|dv0++`5W3yYS5x9DCg
zO#U-lp}!`AMsCQY3jJNs>Y730a-C`V)M&)!^ZWh#cL?v)qRaaR(5YV^@uPbzIu1pl
z<C_)SoKB(4)(mw$RoHz$6BSbeP*xQ|t3Ie=pFs{CZE~X6#xS}hVnu&)4s+mm2^x6T
zpT4f{f$k+4ybMtz`P?CNG0FxhRh4k(9FT3yHgr4+BbV3(IM^<Vl8Zu^DUk!u5NV1!
z8AdzbdBDuyjd?VFl%M+?*;ln`>8ZakE?f*ZO9Lo=$P&oPU%>o}TKF(B3Q^5gwB5#x
zyxKVfJ+~ji>m*TL;t45BcCEAyfuLT21dj*d%hEZJ+FXZKPO_W@_>Pr!ojCnsB$Pwd
zLHk^w`tcoRxyRCHeQBx)--a*ldl45Rg?pvLC}j6wdedEjs{t95&ik^*nL?x#mPvI*
zLz($f1wq>lnCfwU$wC)R1_P*%c%MCYD{ftmMx$dag>x>;CRmz|mkopZKs^XHT3}4)
zHhBBE(D31BVG}J)b3~Uxvvdd!M({o*L7gJ!n$qji94uO1fR4}Nh)w29kZTFH+!RK@
zyA<p{u?pX%-Xl|r`N<JF_>shJxd0E6YLce)ltS*3@1P+hhtTys?DZK*#-qQYX=^Sb
zBwpjeTvd8v<xNxh9Hf@bG#q{4O#z1-Y1=(j6n~JQr^W0+7bwBq%bC2*v~k+60F5pM
z2sjsuS4Z9;xAO!(+a83sc_|h?v4lyV7H+<_CcifYC_XPtmrZifa%?f4XR1)<d`W~o
zE{AbDd+^?UfZ@4B+VN9|mUy*7xp_M%$qi#Qx#w_<Jx^7(w7EqLvENT&{=GUp){vsD
zi=M$=jM=u){b^W%H8r!7?)`XWW`p}vTX-Z2!n<&xp6AeOkKl?a^SzHZ;ckK`jNeLA
z#>S;EOY@~QF@N!KxCND$-ocww;mju43pLjps5q_($Lor;*fR@*V}4@p+#ot9?+pbX
z_O}`jqd97>w0Dd=_J1)Yhf;A!70TgM<T)rA2jOIxG<vp)k%5LIt<V|@!-hi0EV|F#
zFEiTK#94z3N$!D*qk?l$386_y6uyg$aZ#v?-huP?94RtM7<x4`@GNsQX@0$l9WT$r
zeyJ=(cd_3p=r9W3#KMEwmaUTp;_CrhuqFsoJr$5DBZZSw`E@xZg?;vl+^3R+jEy7)
zjx2}NU^&w0<etwx7sNZ(U~j)G*f`6M+GMOza?6OcuV&GqWOdr`^9>#i3#6A~#}O?$
zg2FWJ;`VZR_6S&0U3V~vxyHd~v=+=b12TDmBv~A?<PN_JDl2AD?_ZJ#5#bKS681~a
zGNXA(+;Q<93=u0YWW5$ey&n5E8&9CoG6rjpU&b1fU9dTF6OX>AKy~Xb-0C$U^ZQL$
zKJYHS&h<o8<P-#@n4)}3GWog2z}H%W!g@aOI%E;-z4=^CjS!ypa`wXC58uv7;>{aV
zG6_7)Y^6rTztbR^cp8bu3H15C6h$w#z_4ACSh%GLONgDWgI43)JZ9T{lBc<4yP=uw
z2mMf8#E8U_?~5X|BwR%4_Y8Wya3l@k<M9XQ)<e5H4^>whQ5Io>kO|Vr)>(|{-~DMQ
z?^|{*P{G=Dj<oh`Jaq?&!C~`fI6PyP<T(p!e&mCD%cZfpeKhil29a{)7}PvqpVLxz
z=5U8V>9ZzU8U*xlmI6NTxxaf4b&)N-1aFepBUEfN^v9V~+gxo}cg0Y=#9}On)`rxy
za-1H;T+L1XG=4z~B5c($NAU)(6yAoBtR^kpyA)SnL{ZKyb_!G{<J%G;m<M@~NB_&{
zdg+X3RwGC?Z57fz^{DB%ErQgQq1d$+zU<J-syK?ZO;LEiDh`DTwTNzaBTHr=8d@KN
z(j<174|@vfZdL3^7e?CA1E?A|1Wwa^NL<B(#)`*M;KE3XoGb|^X<<x#t4>C#y0n`6
z5;NO)PT?uRnN;@JskTF2BOCY2eJS#S64V9ZAV*(JxRFa9`hgIPOC@r@jnFlTnA*o(
zY`a#t?FA|_++d#RPvZ)PVngXzB%6wIrg<=R>qe3IV_sKg-ashxTbI38!(inobVi0!
zZh8UatcFnZMoauUGnq_P$I<&P3DS%=rc|vwaP5wgo7+$rWQ39HbTKzm81eJQQq;T%
z8nLP$?SD9m{AJf;lb8?f$<n8+lvTKByBO~ieqp-TJ>*9<qH{?M_pr2az&ez&cs`}T
z=SGA=IZxi;PWbCOyqoT$YFz`KnV-YL=jL>4k1$NmOQY@2JNP_B67|;&k!CGT8!NRD
z|BJgmV~e1ZG=f44P0&wh6ZbQ8k?=;D265h#COaaoI)Q%07bB$IgA`m>puEhKtR9|1
z^HV8WIbk@BIO9%XK7M2{qZ@W*%#L}QjliYFNZz*-PX5fqe<wqy_*hl^40DGH@8GWk
zMkw`)27fH*=VcXolsFr6A54PF_Ypj2xr*#05iEb1L1RXQlJi#=R2vnbb`?A4^&HS}
z<qCV3PN08e70#5X(fcoY=qa&<a9KKzn^rQ*G#}$UhtrW#_Tsu~0_9>fbg3`ZY`O$r
zp-J@7@;infjHGbR#eV-dhUS`8LbOebHV#+hZ>ww!-R@11ttF^thpq8jMY7ZMq6VJJ
z$XvX{uBO$H3ZFu+vIdZSjW||q9Y7wNGO(ys0Z-LsQ9e|JJTCU9S<g9Z;3P*Msy)c;
z+#q!2>_u-Q(ZH{3ux8vw?9`iy;4h+R*(8aKKepn1=Wg8QbDNsJL_huF=+wI%yg!sg
z=`+r;#XHi}W!<P)u14AR?MVLNNPmp@1HB=l*tJL$nlh8=j=CKMF0I6~gUzUw4xmTc
z>>ziY4nGYK5;;ROwJVF_@;jm5Dn?5!O-Ssb7#<B7OfxlSP*UqLRK0x)pR*oxLSqs-
zy4?_GD?zS9%y4L~Gl{*skBpFJ<R5jR%@+@0ae^?CFNWjQ-7#eJ$ea$iXu>&g0DX88
zgXz<U(jyZgC}_Pzj(#tD8q8_$QDFoHs!(-npWxL=Cm4PfLQZouX}3xuYl<}X-Cc_J
zM`U0U=1XUrv`KIMO02v60?BD7Ao7BFqnrA{$gNM{jWpO6a+ZSEz`7k~p>V=P1in0h
z!8fF+FnknLZH^<yR*Ss-(~&ww9G||4L5TN_zIXPZl5;eV?`(wLCkdJ{BOjT3Uaq#;
zo9fChV$jPEC>`#Cd%~UgwU&=Rx)f>sc@J7@(2w+PvkQ#pTJA=XyuQah%@Kw;x$qS>
z#aqK@@dV~9>7b}|DmjgM3o&<higt*iqe?!MyWEq;#s^dN9A$__htiZeYLwX^M;6@i
z65AFA$H+uJCsmJ~om%*4H<6Y|PbK;C3$QzJ897a!7_U;u{?uXgSLRYY2s+4_V)m<=
zU4cSl9%fXmN7#>6xcjP-swMMF+9&e*>~K(-C++)Yj(bW6@pI-H$jGUaL!=H8LiRHM
z@&e4IlBvXNKU|kpqdt8k*)G$<<5h&pXcOEUDu%wy1Mt3YNw<DQQ-_)mG@?hLVv7;G
z5!}c-g8N0He$;Zu7iAAru)OCmV7U^iE=Jr9o<`0koYAf1OiEM<tgCgAA`m4tSI+7e
zNusmE4<ket;9;H*2@;~=dPx$muMMFWFP!Pn`$T$DwHkH1im+s-2sVm~pyx&y(z3NM
z_j(*QNL)q*&yTd9PNA^VJRjJ!0ygn7Q1z0+xd#g%`b33Z%I$!#s2g2kSEg5p33hqh
z0TyVmYhDF6Yz^q{;}9BGl0v1&YG8du5xbAIK&MZJ%EP%Y>ZplXr{vL4DMa(5n0e^`
z5tiPv(6jD9+NfIiFP};8WA{U&w*>9uU!%cdJ_h6|(&e?1w78Plsn1g|^Zsm1lpI9C
z9=(EhB5$GgY&@LLY(uz^7rT}s(f<6GKzW!OrD|-2M?UA;TLSnPN`pp-Zp9S^4LpA>
zM$r~S=*H|2+E@GtOLfD^KR^aUcrIbFZZB4tnb4!J50Le2H?Gb1AiH1O4<COC9)=1$
z7ut?oI~nH5rQn;@XcA+0%uhpk8h1R9`lY8)_G5LrHqwz=-h?1X%99ihuVe0gIF_Z1
zCYuxc5&cpcXX<xB=7Jtpvs2Zg$^>1^-(Hse9zu#T=r1QsXD{WTXha#Z>eK0Ha2oX{
z%22<h{jm4kBJ3U1C%B)&d2osMXpHnjKgD9a<NU>@FV?)?QHD|FhoSd;H7<nTf>D78
zNu<uj!CjKnKgAwGeHqB=R>Q-`OR#bMEbio5V145RY(5f7btP`l5|zNSx)$sUyojkA
zwMcs5VES-71(PS~Q|;*YkO>t*yS_4Bp0Y>cUy4wk?1kH}jcN3;W61a=N7lhU=yDy7
z1sB_46{d(IR%PhAK8o6uRnT*~5UTSAk)3n~*_+<OFta2?Xy_yNums&Ww+;i-U2t%=
zDAFTD&>y;ZnqWh^UH7o!InlKUW4fkqj7_W7;ME#Sr2N4Q6cx@09`~Y((XzC1vnoB(
zk0IggPHee89vPB$B&6z14~jnG!4E5nefkiduGi7*DU0TkB{<tLl<H@S;`62|YzaDx
zj>ju8dqAJyWtl0Tqv;duG&4r{CT2hAb3c8E7ks<>1P@nyf&_DkV>R2cM&Tj)*Sb=W
zkqAsUKcS{$P4cVjG3=l+ELNIwuFsFI#@mvE=2T=Q@!Gs-pFl^k0-9@ILeEYJ%F;P#
zJFAM!I}1?xVjR3rT|n2N4Vb^(j&3>a!le4)&_3}5i_<|(0Y`AvEfybT$3jcJ7&Emz
z=$dl~)ylfiXLWh#Ypuk$v-*f@3#X15O(+#T1brjUeGPpE#d8D6eV_w-EKDK4YC470
z2VnQ4p|mnmg|7Y7g&xmCbO*Qq`6{?QemvC129n2S69{*JN|_h(<YO=yt6$(wyCQi`
z&myT6iBx~Q9VH7r=ybUO`PUR8z4RjP4GyLAKefpAX#*~q2b1E1OPI~aP<780G1pua
zdX>o}df1rErd>uuohG#zB;i-02b9<AL*S5)OHMj)))7U-d{vZ6+tI_yXqdkpj#>A9
zK&bvIJnyqpq1PQ3MSBFBcDYe+QZ%`q`vd39Qb;66h&ky;QPSgs=_4m#1t}vug}D}~
z%<#G-PM+@~=)w*;TvD8YGqqtfN}?H3YW;9ZoU>ASX7tA$F%-`_g@e}<DL-C@)KwDb
zUE?-LNb+l#@B*nBj+9X@fsL_wXu8@4p`Y4xV_7U^FP10AV&;*XamH-50va-HkXFd&
z%6AE)1~Qa=q68&B^k`4|4y@g{6LS`MBKCDZ`YG1~`(|bO{<8wBR+&N~<txrK-i3?&
z9SjILhy#{O*v%n_%_<5=yE1_GmrbVNNM?3j_eT2pJoM@pKwxhMnJor%al`-&I5QoK
z4`$)=yZ+F0U{7PcGHtdVMlGj!o%oJ3-nrFbwv7xl8hp`I%^cESwb&7&M-Mq$q9qcA
zlTLoPUpb!S*xe8-d=|rgq(GB-S0^*~VRLvOZcQ?w$+t4`xM&LV#C0+Lst-+O_QdD_
zNh#Oc;qj}k2n+qbjt&qN6B3e=68gU`1B3<(Wo5*rW=>1VPROu}kC`?-D{Xp~HQ&fg
znwIM8X($x&|GwDE{JzKg+b53^;d}Uf{on7=Ovv~j@6jqgE-TJDGa(}>E+r`^E{pFy
zCMn+6)8PN(0oXF!$kD~t$<@Kh$=1Q%&e6%v*2c!h+0NC;#m>pW&e_(<(Z$BW<@cj3
zXk8s-!jJIxb##F6?+5!o&y4xhy`%T@?cZ1J-~aKq%RfC&zaO+!+LV-kAL;a`hJ0b(
z`uElFZ=-bat$!NHYSOffsc~5`|NYGW`+x;_r76Li{=Vw`ZNMPDkAE6)!mO->X&Lbe
z8UH@izrHiH{x;O3|7EEEd}(a{eVl)Ny$AeloaSr7|M-ypx4-RQUJ|pv4HR7c&jbDQ
zn`Lx>goMR^Z?7wA`9CB5e$IYh{?E_5zh4!;eCN@BUwAvo-#@S4_^;c){~aA5{-3`t
N!CU?Q{r~#i{{VRXqJ{tf

literal 0
HcmV?d00001

diff --git a/test/models/data/transolver_irregular_output.pth b/test/models/data/transolver_irregular_output.pth
new file mode 100644
index 0000000000000000000000000000000000000000..8f3d852e6be77c49d46939f84a82c4c60c8188e3
GIT binary patch
literal 199129
zcmbrlc{G(@_&$teO41}nb4in=lsNl3nnWSZG<;|fO-Pz0nMpEa9+H_v35Bz7LQ+YR
z1|>v^2GKm~J>U2D=evHt^{)4Q)_T@+)>->J&wcK-pL<{TbzR$HML%f?2{}25|Hm;&
z0unCHTOAHK?mcAdY`x3b+1Bph-mT8ojt5<w4!X?RvDIa(meZcSMvL?%HvWG;+@((J
zmslaO-Nn&)yS<k6L6=>74=7kW+ip9!Ywr$g7hCb;ovjb)DD0Qi)U=cp2YEm|a;u%~
zeyRWU69uEaj$2*U{I4$+_V=@t6pzu~FKyXReBJJJaDRWZS7k2yWmc__SRrk;KzoIx
z?|#|;`}#kh<ovg9cig+zcDu_iM~4FnE1h@PI@|78vTM7G!hZRwtM@Co>>sdtzv2Nc
zixskR3-`M#ri<_TKc~h2Hf_@-4F8X5)7t6iyl<<EwK%5-#1A!EB>4Z4y^R}Z|4(Gs
zJ~ny&&;S0P^M4=NWQiI7$H=xFb+L8)p9~l+()@oN)81V!E_-eNCkSuWK2k;;!2g{8
z`yeJrO#43u@jtU_mbTF%jsKSsJlCs5t&s>53>gj;FxF(A4c)p2^t&^Gyq=vxd22Jy
z4V@1D*Oa866k*u9=PYl}VOo5|1X^xqsdnv7XeRHX4c%UFYMNqDD1Qt|dvmdE;$LR6
zG=>C6Woh}x9=wZJrTaT7VR@kxLly3^#EPp((XtXmtu5lyulrEJ$1G^{tH#&$YoPXM
zE>F|#L3q#_+UOrf3#9_l`#BO^xdleeaeVg8!L-jWRxn|9HDYy?nDK!YJl$Z(bH+|Y
zgmxSZE<|C4k|*7<+k|sXGE84%ImwiM!SZlD#F>v4I7oPs?ekE|Z23()m8L^M_Z-eF
z41?X>{q#1ZkFV{&h)gfU!*Hts-*CiRaCo~n{yE;oOhF*q9y16b=Uc%QR?whLrI=Q6
ziAzl8MTc4g5flWpx<+H(>L|YT_F3|)yvL>tu4O}Vq%l?|Trf4al68+;4Z9mRq4sbA
zfgfe-7cd&*>LW_IsEVGgy9D>v<nZ4z4dmZ7201S^;dEmld$D^K?R@D%l`+}^8$F@H
z+tkNolH*NU>L=hW=|ab++YxH|3n!n<q-_6fH1S_KDM`LzCboyjuGXKe==z61z2z9A
z{F5gPac1LHV&M2rnI>7L@p+r0$^PX-ta*8!uPF8*oAV>tMU#K5*ZvdgDs6B#ES;|3
ziV}RkYl;<uP&^NJLrPpRMH?mH`2=Yk%c_99d?D>!7KM#lq6H>%mh#WJ|M*AMo1j(C
ziT(3urr8IX)gyJZ#k7*fwRJG~yn$?eo0;FrK4uhk7QamG;bqi_-Eo6RLbrh$^ZN_^
zOKf@fwj+34sftLYH1PCRl67c-WPF0a>{Wl^ljwg;{+0)N(&ET-iz`O&bAa{c9A3Dn
z8Xc{9Tt{LD@-Dv@UD|StW?%WhZ~pWGy*W+OE_KoSzz8Ozx(V&}>1bFQLh)%AAeVfJ
z9X_N;t>3S5mouADcGN-i;Gl!xM1mTfxOsx`dx&tO*GHaq;3}LOwMb`Gs9<IH86076
z1?~%`QA1?}j~nBGmmQmE?ubH`C)tSdc|jODbt+XOuK2ymCM10JVe>CmvzDd|{CZlB
z0K-xE^q081qFl)yyK*d<8-@Djx9qO@ARc^Ij!NDs;NTihI&7c>?R_mgYj>GQ=gJfI
z;#C@Lc)N&u9KOtL+vh^VBuwBPEW+Z<dg_1aFk1F_Q1i7}yuNP`N&gF@<asmcv%wRX
zX!k2MPY9)?k`BBPT%-Pdxv(lyq^!bK)Va?E2j|~s7Dp4HVONCRR|Z3U(gND_cNO2e
z>ndDZyLexB4$T{PQIKyIj#jNAwr!S-u-^m~QaQ1lsvV@bdWs*lPhZ8`^fu%4jx?I9
zoxqp)xG@)}Idp9N8|s)e7~3D&Qr3$%g6z<K?0)4dVsdiqaC#nJzI`Eue|`-Gy#`Uc
zQ-91#^8?h|U>fy}662I;{a8ui?$0u?ssQi#*JdCp^(1fAW_0T|aMyWfDX{qpb6$J}
zYr`wCr&2&cI+2JEcm(4eJv`@kn?O0ifj9hDiJ4rEH2;l)!%8jg@OD1t^(In7dm(>$
zJRakMH^Z>ERq&tTYG|73vC8TGC=5K!?Us5{cYqFTdne$slsY}0x(E4xr!d`9cKCMj
z0Yx40rHX<WXlFzbt5Jb;Xc~8ttJ4qdKN%ZO_><0jUkDq#sb^wSvChQ%<T0R@idXmX
zPKi1;Xo5m14n4!LkW|Jbr$A-$2#Pr{3cfwRaqgKl-R`=@zr{X>V%sHj1}>$b5)DfG
z&gqZAb&-Lb37#*A$G~?Rsosl|eEcoyU$=}6YzV+w`A0NHs}z^Dj78(_sIiriyK!cE
zH(D3ofTpSsRwtyP&&HTV#9n0D7bY>ct|Th{H<X;eL}2X1h5VZ}XTNmr!zb+$y3V#x
z!+s|Wjx1uI+Ji*TFDkLqi<BtBLO@#|R3YsBY)B~I<qC%{kcY1o+y8Ybi~VIu&X@WL
zcYFOpOT8~^?HWygV!}axqu>@Jhp3fx5RFhpkWVZvGfcwvm`L0S9YMmyPiW3hPv-P|
zEi*iE6xTJb3jF{4Aia_tdZ%;+_YsA0UE{I&SFtF<K~k8Irj7!|q0D3DRCeO|L&1*4
z7iiqO>oBXk3DHbFj8Sl+loKP6r<cLs`-r&l${F1J=xIpUKG3^Gd4jc31}I-sOWPXj
zknvBCE+)=kJ+trN!y-@Qht7c~*AW={$(A<DbkUoc{%G_pV>jHUL*m>Zde6`C;^j+4
zQqm6eqA#BQ*yiGtwI_SL<}8^Uze(RRZ}ZR?ce)<*iaReoj}sR%u_*00JyjS$p8e&7
zLFe_zQOODon{;T@lXsM4mIS}r8|ZfQqi|OhHfnMd%)KkgHu{Xnb;~P(?$QX<oyZZ~
zf3gx&m!8Jsi#Z^hbdkilI(}9;pZP7EB=Rwx%&*I2vA825N}ivKW9#N{!*@YUwptZV
zERe;&yF@$QpGGU|)gKXC4)e>(wB$w=8kNKF;gAlR)W0(CFRN+uZAqc-!xFyxhC6C@
z4HPc7lc#LMUznReN~Ae|3F<FB=IdLMMV9)1Sn$G~h;+RS$JSgJ@AANat>MW0GM1V8
zEaPv67?R?=4}wceBl&{<li+!~fgFPdVT4<c$nv}+jzkV%8y0p7_OA;k6UC9_m>Eus
zp7iJP2^qM)IvF1B4R||wCrKw8(5K_0@ayOg);WC>mEEg`PeDA-{%ni!f@fIk_LbSi
z7qA7V4%3hlZCbAWl)bPC!s6dLY;tJ=(vP3VsR~alX$8iIUqfT)Da<`KSJV)v$$b7b
zlaYQTMm85QqlEWt*Y`)*`6&RAeJ!YdFo19L3&jUfB}O+br>B~Elyawn)t`z*&E+0y
ze)beYUic$HoR^C9L7-V4h@2QNI(FwTnI3eYC(FC|z{g%lO*+G)Y?feiN(J5dJ(G`f
zD~5^T5z<~;jXoEDjEm2JXw+#Ot4wEG3RkhyHrLSIvW>kuH=Tc~K7(gwoy=LFgp-#i
zQIZaDd8I!|cU9w#dodPwT)+<ci;~%eWc8zf4BK2w(l&R{8r{t_%*q4nE-RGEtVp79
zEz4NHga4p{8@M@C8so;E$MvhHXz#l?*k!z9!mGb|vt%6X2dzh5>>9Rs)=V0zx`?aW
ze?(5vP)ajwU^_CMsAcv&N`9I`5y|P$TX_JU9fL_NW*K_@-ox}>8np5`WJle^p>#ts
zZ%C(kAMXfq`bYC&e?tWi2LC{oj+C%Dy#OQrB#=$=3~Z@0K*9Vl#LAwhF5P-~SqG5H
z)MW@fAXB<}bQ~NuJ>sVFL%Bs$7291>CwjQnkOt)~r%l&m>ELp85>Oxe?`s=tI;@6~
z4T;#)6J7F6Dg>STPcSRj^LS7%u8rz7xRhp!r!HTiDy74}H24T6y#7Nw>oZ92#z%fH
zKa4gOkH-B2ZS0|~JJ%jRRZ!@79;Rx#Y=28Hl|8V=#o1nHZG6UZZ=I$+Ctjhf@>%hJ
z!Iz=eCXapDwS3N84;rv#3Z*|w#qQ<uR6BeM8XnlAW9}`A2|Fdonr6k5R_0K_u_S1`
z2<M^IGVsmqhqjlg?Cq{ry!}rGLjwv$TKhS>x@{33d0`+q*oBDfdiqd4U_3n;5=!0^
zwn6JqJqzFblO?{;;?A0TDC2;5UjGokCjuXmWAE7PyQ@)U@dKlG+`!aPF?7i$QDl^6
zf*wl;N?-UA+j^g3zGf>Q;v;~R(r);toum<7<H&thIj8R*Xl>whK21rFKF!S)T={+(
z6=Pf|+btJ2_b+3LrxkJI<3q|evA{*6rC2Tf4l|E7;*QfCcEG`uCLL}=*ZIGYpI(Qv
zi;uy5i!xs&d6N<k)nVLh5hb|2!Iag$6yH#dTTzM3cH=O55gCebtyP@tOcC=|UR1bV
zm+J*SMyj?RG-@3X@jafe*4Cg`ol-P0DOix3K9M!YuSag@LM(VRoTmN~<0zM9c=BT%
z%^k_bINFl^`euS)?F*Rp+5igG%19~Jr1?21Z0YvTTqCWMReCSM<<-5SNvn5|;=ope
znDcM=;POyp?vfBbiv7u&)nl2D&M)jdu^pM8W2oi#BX&q=hfM1^*c>tg0}M3C@*i<w
zNeb=WBo8TL4d$5bN;A*<LGn>4ozZtkRjnm_r-<j(!(Qn8>|t{k7Ga+2Et*lD&qoa3
zN7DvILT#=gIqdes0Lkl==JcK{3Tm(?UX8ZC%B4DuPCl>hmf+IPGMf8SojEmyiVWu#
z^TD-K$!Y2X@a@6SdG-qmQ-5)#@!ib0@H2H^dJ60CDi*)WhE}tSG}5~bO-o$_p|)=Y
zbx}vT7yC@cVd~hWdj~mRC*sxuDGKhMiJg&2=v%jt`pe{E7(O$JXY#nQ<11z4cp~-E
zNPK*$guLgGh)n+v)$>Y8O3xOS9UUdk%O2xWfCdHCWs#Iu0a=XkH?Wt=MaZ58v`)HD
z-pLvGGWRX6X-;7-r^Z2|ph}P!rNLj6d(t?c``l6^8Y_=}V_!!mBlq=BW_71O-#cO+
z?Ht)(psl-!Uh3_HRh2pB=~{D<l#Fo4n#-g;*nkE8sDohe9Sl8uh<sBxqGOV&c6E}V
zYtc4zu0F>TRvFQ`u-)`{P9loCs?eSH4(ci=O4NrHp+IUj?&a6=$;$KL-n+c`%k5}%
z9w;H{37?U;B#YvkZm{_+{Sb9`6YqC7i-r}hM@Lo{+D14r>m?VkS#SlT9x+Uk65#Fv
zU-o3kH~wQrlEFfy>4Z~FWOca!j=e_MEUx{X^#iFnx`gSpSMbI^x%9`v7IPd0^iTaa
zj4EHUd4(bzA2bHGPqVQ7$~mkT$2Y342bT7g21&#Iqksn=*nXYSf<Cnp-niEsu8+b{
zQ*Xw0$6BJvu!kSr>xD0OI_OilWGS7mU{43B8??T>hdD}B6kV8!YZ@NZ8`DIImJ3SS
z{)oqho#B3Y3Aq0~mD^k1#ibo0Y<7rd5tA5i-*_3Tb0meIdS!&l?oVL&cLK|^9|iwK
z8Ps<+oX+)Kfm|~|M*2CM?QTccx~H;-0y7*P-h}KSK_b<}zdXZtu4u^5OLW%Ukmlw3
zK<54|cs@yB8Xb>CV~;lBq_qL<Z5>VKL${()#|vet3dPB2XLiD=ST@vy|LG3K2QkJQ
zZE&AE)c@ew*EYlAxH0Re_E*p?|BIZ;yWrw+g<Dq!@`L3ue2&KnETO$fsCrC}2}N9_
z=F8jHO5woQSM<ouh@FtW16|Q<GBQphuf{CwsmP>p!YF3?@F%sqxpOa*KTJXM9QqwG
zA_-Q6BkD7#W~wZyTpL6`0z9EIa1jRocBdEX?{lY^<y0_q2(NY6MV7OC=<?6eu<ldB
zaO(n?D)sO?{*RcZM-gi~Y)1VTpF^4JnUZ<Ot3{jEZGo0wxZs!mAnuYeh)FGPqVq#%
zk)ov+sxw~jg_$3?R(&JeHuW)jV!TOsZX&JKIS;=@wV1x37=OEpP(H1e^*1KL^hQ;x
z?CIo1TC>^9{vmXE$rre}zof!3CH%V49+5`991I?);8%zhzf-<}6lX;6!>ix$5oCqC
zJBG8gdwE!0{|)^gDzg<&Vo7DiJN9^*19f(Mh1!yIob{fBVfw-J^pzuz^7@CnOdG+c
zY*5Bx5o_?4pG!W7-4j*e>f%V36QbA<uT5w(D5WV$8?nt}3IqZxrt;@9j3b5${C*nq
zhxa>h@a$18VfUK3*c`y>u?>8j=WYCm-M~uI6ydSK0XBL4DdEpO(z_?FB@<rK{VnHN
z!G>Ej^8Q;sx8e*ma}L8TZ7Fk6^kSoqt;Y(VG)m34gH3fCVs{-t*Pa%nUU<xsE*3y*
z`e}+EpFn<{@-*<d2X%Z3g7>^+%-WleWwMgOzOOg<WZi)@L{SN4i?T>@*;QB!>Eihc
z0rcU`eCR#;iZ53W(4U2g7?~J^*G2Negsp#g!<sz7&^O~T)lVIJGsd#_A;o-HRRv$y
zoeoRgBiO0gDmdh%NP}LCB@O#BW|h;1v;TBx%#BdKd07IENTrgky(Df{-^IWKV|YPU
zGb)dcq)~Co@QaH?zdx6_*`_SGe|o}FKc!Ps{x5E{sRr5Ui%QHZi(r302A9<PaP>qD
zR}YaDs&2atJ+;4lc6$o5_^nbh=EGlNiSDFnV}+;d*I}o{VEXlBBc*JJq_W5|QK-fY
z^k3N@+n!33YCr~0byFxE9<mq-2M5FBpK_@|whA-?&+*pvvgE$3omRLfV3J8W|9E8}
zbAP*s-)m}srFH_<E4A}}`}af1C{tv8a|r$$m_R=a&WQWb!9uNZ(YPYHoE`aentp^f
z@$_Q@aIv)r)mgV-YEVUkM#YnTo(4Df4`+>0Yj{sdG~3ZsiKIoTtTpXAqTCWuqwyJ+
ze_P;1+!=8$z6y?cO9<zf<U!Ha2z#D7Q}>#17BYJ=AMt)EMJ@Dz%N!XhR(J_V&tp{C
znocriQ`r13uaP=j+@rMna=~mheBRcI>}$Vq<+2W$Oe#aT@@tIfn1{w0Z>i&)CQ40K
zq3qUlHe-M*=GvZuO4K<zn>~|sXDgFZ>}?wI=?jbARK~U*7gKhT6Uk^|4~v)G2+_Jh
zf=B6d@lE?8(^>ikgCr-?N?`$vj_u*vb|He$^S7}!qYdAGou>I}UX--`EekQTB^|>e
zGEI=jONHkY-1Qt0ilJ=9gR{hSJMdysDt^1h;QHdjczaPo_$OS}V5+1WlNsp%&xIn4
zZ2HPpzrD}=S^{Zu#B{{{iKQQ5ee6c>DYOcG=t-nI9Zd>?>C<GA?o+}1Rc*MG=f%cP
z&*P6}Pr%09jjSIuvY_5pQBp-T<<*_1!3~^77~aE=Nzy#S@g-Z00OXWTq{5(RG$@Q^
zCFwfku<QrEJspEuPKa<ALESISkg!5dXnCU<$J|4i&Y=iC`eXu{6|~rrcZve%-!g<1
z8K{YO#hP|K*4ggKo75kc48OY!9rGoH(e?VICpSPOlYIw%_5n0PRh~sA9HG|FJ_gE)
z+4z>^$C*_z39aXmX8c3Je_4a@&LV=}Pdr5fqPtlABiYjOuoNzN$O+QZb4pGK$CIaX
z9VCtyW4P!M!rrZ?^M8~?+CwLj*`qdgee5V`YU<Ffo-0gVq7T~P<LI>SHr5(AgWJE5
z7MRbTffG!PK8@(Y0<WXcbe+m7qPkE%ZU@$ddqHHp6(_{K!Z5u>2CdPO{8DKpmwP+~
zHAm+OX5CpulUxg^FY^g$pTAD=34_U9v_iikAQz8X2GXk8C%D3q7OZbBgCVz}Ir7SE
z_@`VX75m}!#toD=`UDF-I0Qo@z92(!3b}fT*h4EZ-~Ff_RZCTv#gX4wwZM=oyH>$H
zBpI{seZ>UI#pF7B{&JDKsO6Cb{1nddb-_V6GEu~*-mim4aVFM`l|gIxJ~mpR5T7q6
z@y=@!!gq<6FzSI69hZ*egDnry<QsX|{c$^<KY4`Z))%3amr2temC?K_p5#^Q0_nrm
zXuY}HpyBUEZqv9x^vON|iw4?|x{WasPrK3lJsaRNe3($?%z9D@O{To*hPd`2iY0g^
zlze=tfJN0uN%^8BEZ;0-^>;rq^N%Wc=Ie=d-Q`sBG@j}%JcslyAHny^uO#PI&s0bK
zM|UUqllHB#2F;#DI4&Fxsrs9k5Zzy>vv~(hH=h(mxSSQFTr0zdj58Rc+fCIf9tPK*
zwIM3omrhEZqC9ba3Z45nhU`G0PJrlGPAQh8_rukyu~aWBSE}IePov|~aQX6RgXi(`
zv|sN$+<v<A+#z)+|7L{6PJdD6E6FYTmJ5Dme-zERuL+eR6LhI1@y}KVxU<$oSQZ5^
zv%@Zs+1iQfs(ifrkW42E3HpZj1>e8i!N{^i>dXtk;ILF`lX<}&?EQf9WkFoE{2H)o
zGF^T-3JZd~_yD5{d>QG@XC17kg98S^CQh<cv}G=-WR)OoR}po6&4=^)fdan;6-?sq
zZGq)~{urP?9!+yY;TOCN^$HR~kx38A&W(rZz5sgUw+_L&0m$0p0?(+6IAQAtO-l<*
zTX6(e8;sfSA_p2~cNu=lvh>$kR``DLb!=?AMma_YSj;6uteo9MHRiv#tC;VRo2v!Q
zDL0_HL!BQPx(@vg%i(k6MoKRn0?`Y_QvC&uFkNIpQ&Obp)20v>BfXlNC6r><>8+ws
zS7Sxbu6jcwF`Pd?WlX2V)St$8Q!3bV6`R6N<8Jt8K20MW^J~YF;)qGO>@Zc}-rb6t
zK1Z75bb*F>ec->JRG^-^*^33b*eH+_8XKf?qxDm0MaN7MyxI<vO`UAf{te`MHwI?o
zlX+L0bm{75dG5Qr5m^(8S<8(=cz$07uXY`@4l;uE^I*2cAQa>k36-h6a7)xeTtPL|
zPNmVL&@xEMYLKUt9sS;wLTy%@ZA$NEj<+^brrR$;=8>D^^EQ+P$&24VJIT^;^V2Nr
z-bC{6$!8-SZ}Afu`4HtEz>GnsXyns%B)jfi$s|D>{gmWP#bGBN`(H&+kp=FJ+yEuB
z2JX@4gYo@`@QzkR9-4U;j{3Q%HQq;}^kM}3QYSBACQtDA#4beVLU)=Ew8?=T6x}E7
zRnoZODUTOULCBP;6(!o|^6oQU=o^-VJu|H7z`m!vd2=^r%-%z59~&X~_D70eZb|7c
zSJ966olNS_eRgcO6bATI!ue*Q;JW2|G1`8_KD(d6yY)gIyr+R{cJ30yd7onP3sd?0
z;W2bzRv5~kN<qf;12(8sLF&66+j?~lw5QY-x2MaNMtx8usrh-_N?|UQ#EWR|2@ATu
z=NkrEyO8D8Q@r?@BE@Iuqw<3XRkuvQ*Ki?veh(K#RXoO+73nOgAx9KaKZ!lgPejz%
zmn>$!9YY%DX)=a1{i-Vk?H`EmNmJ?A$A<{au@KbW+r@D-fwK20Q9}Dh*syg5!>_Ny
zoNH@vaPuAZJ6D!}4&TaBAEZLjTv9kRISvP{6QT3)11as1AUpjux)EF^SpUHi2}V0n
z`T9QIOnZq5o8BSUI+ETeMT*v{bkMQZPUw4=vHr&-p+8XvuC;Fwpp_}OT;|C-Po1EZ
z7gLy<sF1&(cuKJO*(<z{Dn@T*8_I`D3&GCg%f{DKG)ESG*#f4PEkx(TjaWWX2k!eu
z!EZ??E%6WIs+v*onRE^5*N4DJVlMSAbVE({YB*h42Zu!qF<{Ag^o;YvlCi}!Px~kC
zRH;EtCz)wxmkZp@ICbPhP`^o8=uo{yplhkl_8+#ScP77B^eS(_NtNWUY@kWvIoA^@
zFTB!wg}I$=ghY8Pmfkg^B&`r~cr`>QGW$g<jl_JS;3{;!-bZ<E6ijtK!)|XZIzmHH
zIQB8V-*`ak@)49B(1>s9N~J!X0{Y`3!I?xATl@Ns$b6C&q>93DMYoMLn=i$<(J4%8
zj%4YDX9als%>v=!CKNnA9J3Dy`Mjn${OLG_Pm{*em|@;1oL0c8@IgsdV74HvM?yR=
zcZe<w&g35VJ~Er_arm;_6S5mknPr(N6muWrsP`?_s{IS^J!XKUT(Q@xgSD1t6P>(3
z{&6bQ;Z=&fV^>Mxt_t%x<;RwFq=4tV#LnkUB(bkj|AXsC@|2uOanIJ0OKLo0b3fvs
zsW(P;+EK=JcTtAF4!^Im2zRUo(8K0aq~$t_x+SMUU$z*P@(=Ll?`CdysS=^<ley!z
z33T+}3+R1O!~3AY<YcPDQqtb>nE@Wyp^y&qW3f2>`zoYA#PHH4#@)83^Qa*P;(EJ>
z^&7tu&7;5IP}V%!oa05RT|w+`g*g<y+#qEIA=jT<hKjwMC#$Psy?zO9oEr#t+x!wL
zyMQ@&zp@~;iIg9`020R{O0qKT5Yhh-<)nT_!@0NY(`GxEjjk6&MpnY!xD2C`hX@P$
z9zZQVh=PW$K#Y%EX-(oadKfthF;9MwSLQ_sTNFyUR-1Ue8X)dxuA=2d5W1`Ca3m+3
z`JV}*q^nxAs80*ur+Xl##U6QIdfA<sCG2Ww0oc5o=u8R0O^>ZS@q8FN`FB2IPn04!
zT7t$*zl@%q2>l%^w<AwjD>zf(4F|JoO6vH?#>9*Ju?E7lnL*?oeh_*W2C~BHD&$wy
z2vh@&Vc~O3-z@q%0`}G8&!;nZRUQHL^9&Zzp%mEl03UrD5NBnGT@ydi5RGZ<VrC#D
zmfPqn{(48MxX0yB24I|OECN&;Ans3ihx9)FLe-G3yVOgEQkRosm^3TQUWkx$gGA%=
zqM4#Y6lA)S@JXo!?Or$8s<j`fEqoj|4!S}^ho9oR`}eYmb@BA8w~iJcorJj7B#bd!
z%5sjJC1}c%#w2CIzS>waTo40;mE(C`U=qg3hhpWk_x#l@KimrMU#cf{72CH(Q0tj9
z&=@D69mfhWD!Lx3Wd)enR4Te~s<&jK`wIF#UXl)Ne9p|zR*-$XHDy0LOwD(tO9#Dk
zM&j$&)Y{OWbdQhYsxOmBv+gN;XaS|nxQPDe{;+YoQ^@qi6rR2Q4eyxU4<UXV5gN7=
z`Abvq{aUzKgD@6&`FQd?rK?n(;0!B@6!pbBLNFi{Rau|-VTG^MGsA%Iy><|vW>16C
z`ZqXRoPx)5zKNbp%@s7gIEOsV5q$5U(O44Zfs=AMNH6PU>M`kfud*Jy9p;j8{?Fp1
zVnqzuG8it#V$NCNDmu%(@i1o*PnuN4b*B1@_0BIMsZ|59_@oydUKfr81?ke6I*ajQ
zvj^Eq#8BA$41wc>aF#EKgX*jcSnX{=kDtHiF71_6yu^jNuZg`N-pBFs-wON<Z-cRt
zAHv9*0@s#eVe>z>Jn%be=6=PmDo+YD{>rs06-y8652v{GTcOv}gBR1{nEt-c+$r=C
zEcX8p)m;0+g6ywB`Jfi%r`JROp$Z%>d-FB(W+Cm4783ss(@$IKK<{#gaw-2TRP|WG
z{lA5jWYJ9<&dShMs}T73t$^2_zkIXoOzJNePtmG@7_k2WIls7t<n7tC(CH=xuF{~f
zj;B!-FU0QBwKT0zg1snS!vAw#jGZbEMBkl9LZeq7;iK9`$8Jh8hr>NMa<hTwT?;1T
zie+@L&mZ!RQT)_BEy|A8$6}35q!w9*wQ8T>G-D)0f9CPNHSH+-9AC02X`pb>+-G?3
zt%8*gen~2qPSQs~1Y$27Wx`CbDC2cp>d&7NS?kRxt=vvO8rx}ig*K^XuO%lrU)nQZ
zF-{E_4xQ*Th?%mH7me{kO?wzrea_OCjdG$RS8h|Ui#=Opw~vp!d6vE&ZO0CaC)77v
zpL!d2AWSNlnJk`#d~ZKAp$ntff4um~FW9=opk{0!OQaIEL^+(C5+~tDQ6lLUT!2KP
zH)-6mquv7_vFy4zqWupcMe;Sd{^&1g7x%gA%lFevx&7ET_aJ>0<8QSHu|}8Zi`E?+
z<1hRmok!yS#p)l`1hlgW*=nre=MCz9vx(d`xAN5oOt8$c3RTtBFfk6NS8jDoGbsV*
ziqc{7^ft^sxErL#`e0b<HO%(PW|l^!=;?h&3d5eE_16RL(Q^eS7HiOtns5BBnGaV@
z?#4-`0l}IQJb)`cG7GeP>_^taYT&u?I@>5yiCKCwT)H}#{WZMJ3Lg~X^33IY+|0rF
zn{@>{UrCkT*Ox=an6dcy<Pv6z@%EB&m*5lZ#H(LL;KkDQ?76{Byiz@cpHDd5Ug(1-
z11->L6KkOS>IA-t{YWb9TQEg-3j_tlI3Hbuvr4V-+xwO)4_P8=xLbs$BQnsVbC)lk
zCCiT&Zo-{6(davV5k?`A*j09&_pEuukBlFUA3yJiHnvPhZJq(-J)@vAa0uJj7Ymi2
zmH2E|h1lH@kXmek6wg|ublD3mm*1qeekJ^(Mm!686T)@hEhg`k1$gx68G5p_#c|Dq
zny)XNTRIWTe{yE;Vl7%{o(@`i4)&`%;nMe*X(^kNtDFG^Mcd=o{1CzL`O<WFPC0JP
zjHa2>-FaT`0t6@zqt-kd-u-%<D0`n6d}y4d_4Cc4zJCZ@mb?}Fcrx+L+=Qz{>=tXJ
zMReXh6LXE0;m#I+))ji1#{9kszat6;KSoa$^-C`iwIYH#w_fHETc4rXSdAh?Wwhtq
zU<9SD<?-jO>B%)`^qPiZ$j)!vv@HcY?!8AwupQ2}U1oP0+ZcPO%gy|ygo3LVc=?1p
z(bIPx=$*3_)vZ<7Gb<ZTWwK}%YaB=452wpxdy4nxeiHbqogmL1MFjU7DH^j`l1d(k
z@$NTGDtsqTgFe<EC77{ml4+=UX^*d7{?JP`!~8cQdh>WBZGAofJ}VQj+_j2#L~Un&
z^*dm4XFo}$$r9h-#h0wC<jK+wIJo*YeQzO5O}InnzlGpLTq1oL>p?v}dZexPn?lb9
zQC7K_TU)S(KYVkFOy3QG<KHA$*IYz<&sxlVVL-~e;&C%!K4q9(p{jPN((4XE%to@R
zIC*L?eA=?P*>+#3#%MuO>nQS5LIiu)GIY$2qc1ad@P-kId~lo^7M-jRM7<bMazbze
zFO=&_w4RCQx7`)KLbVSH0lUd~WG0P#H4?IKrI0oB3p}DEg{`u0Sn0EWe2c<$W}2bM
zZ*KF&&TajKO`37E@QNWlktk*3-tUI?hkDj<f3V2)q8XV-7L%M^Ju`{kg}EbMvFwwD
z7?ATB9@CZZ_(wLMpZkTsW4<t{FCdu({e;2`U+9_DQLe0)5986Lf}N>O@Nf*Ji&qIc
zMuSQx82Yo5uM_C*w7XPOya;*q_u(va7urj+*u<stx#FI~pcQABi~lo$eEk(c_2wU<
zR`+gJ!Xl}u|4fW7EMexsg{;oE7+=yB;C|vE=GC=@K5rY&En5w#yWN|w+*LwDHFa^V
zqL)Qj+~%pPM0kG82~F2e@`+RW36r0OAwPU7&(9e}FEd7w*_OF@^;AN5Y+yY9Wxhis
z15+Ay=mKSo3PzOV77Wy{#f#z}sMjf@0}@y0N6CG9Ad^RLJX^41Vg~k(93~QtnMkc_
z-wkYbccHXanU&rzpjWH%ptrUNIV+|i)LnumFPx0rh&nu5-~&fj5x=f138|o7c+9e=
z)^|y;uzAJOGbUnMjaWx|Iu5V4zUB9=!f9N;I?>Dr&-tW5BBnkz7VZ<yBXjjHruye5
zj+M+6>z2m!<y;8H4--;HX({)=wHW!&yabOce0iJIFZQR1_&WO#uI4<1mOg!fD}KdX
zV|g@9JA6sBe}p0JwC%&%V^z5E{v2NU<p>tptwm^p1by^a!m7X7l`PK6XDa8Kp)qU}
z^1kcR*wu$9bom_yhaQxK*04oA0+NY1M@_c_aXrb5w0D|ewoxj!w^uWFk4YuNxA{<K
zBLw*mhU5OP60p0H!XJZ@(R*wfz0Ez1j3<#OTbqhkKhF@#Z?YkR$4oCb9A~7ii?&sZ
z@txJml0L8DJVreJ>D4sB^VtvCUlTWKnouDyo&5xBVoXr8sSfD;!;h`+#YF2Za4Xov
z25!0}3Y@)xI!;-0k4$;UjmxE<Q*WT$&XvMyw6Lx9B%*(c8@xY}*qXi(`^Md%6+e9N
z-^FUkPOnAJnG~!zl8+X#$D_ERM09=DP9||g*I>*oO>S3l5zhxF!Q;Yfkx|zh(#|@F
zsLNu{kZUt0ZrFqD#&vYo;xy%nHE8uB8Jf^P5au6#qaxi00h&dSYq)7JWB(zfty2|s
z`Y*=H$=5LTy^K(OWIBxNbjfMjH3+N@!ZXO4_SJUtiYjw5KkI{PiJ#d0Aqb+7OQ;od
z{GDdvnqImMCC~3M=i~+K+sDy7Gd~#(#T)VE$$4t^IEynI!f3X23$sjdrB1~`czTAD
zQtA#~_317xH5B^@<UN>c@O4z*4q|=d?(sK4<Jh3W22#Hn%7jOi=&9Q>u`an1eNt2D
z)UgBDaApu{E`4FWKku>S$CUV&H-@xiU;z?42IG~*94g&43)On3sB`jKgEJNP_@VhZ
z{K%66(dXrBAbao>+$Gjg*BY_*w{SFh>5j#=h%kDpKZ!Jx2U6wvm&jT@9SK1{5#oJc
zuwPAvCVq=%o@_hvZG!1%MlR$6GNBs0oiFJeOEuQ15cPH7%m5+ws(urV&vJnFs^{#L
zdIp{?uOXY`6L2ImfN!qLN8%_&QjIyt*FLpnj}JZL!4J~tvGsY_+kIuydKU!mf6hkR
zog%DtzbN(`-eZa<tGH2`*hg)cEY>5#DR6KItuz|~jqK^Ty!{q5yy68W^F+AQ@|KgS
z3boxlg98a_$Z$-e)|ubgrTHJ>+i`*gHasnfo5OJLr=0M?uw`s9NeEZ(evFtJO|&kx
z1zTlG%{Pmw*Ow@-!H;^2P580p9jyLpDt;P&p?s%%*l<4=H)1PA<&RTvxni>5s(&xD
zYz#Db+<KqwDL6|WW`5}Ezk=y$k0M)}1lUSn;rX8~;@E|GwCVg3-cc*|euQUGZCORh
zG3rO2)9m?}ur&U$%$<&3YNgZbGZEDO8mIKnk@1`lXgjSW$eNyxEe1pR{k%kx;pVLZ
zhcbO^{Bwsb+nk73g`d%!K%)8oWN~E8I)nC?^AS8#?4Mb?6>q=$QP1#0IAV2%AOAd(
zzTI-9ft7E0=o=H>Pd*(nhfYFa*CZ&97{eX}Pr}DN<@hm1yTnW?2E+bN$1^({6s!5e
z`Se3hEj1XosfaG9hu}zEDK)bSti2}Y40_!Kqslof{4SRqpW4Q>PP~E7Np)6RBVY;i
zkeNU5rQHUq7<ROn&xr37M6Wo2V^^}UYE2tLYWLEiOBZl9;S#0CCeU9aKh`zvK5ZM3
z1yj#_t{0xi8owEk+p{m6#+l%pq#k;MZ?U{*C-7vHCsir7V_I@De^|ei=#4MNju+Rw
zhQEC0wmjU<Q@}87v2J$l2u!#hUOSZIYxP+i&GJO(n}rk+SB--keQ5L08MMMGK#bFT
zxKw32{_bxlhZ|Xl(Z7qvQL;iCg=bvP+m@Cn#F41?9p)8?V0bJU<>4M=yds&-tE=HU
zf6kt?IAN$qD!xaaqG`XD<LCVMqOvI=xbx2sPSIyEvRxOu&F&K){s89^E1>^345=UA
zl74C#K80T=o$Nw1EqE^K7}rJ45@mSR?njGFV=1rT66IJmlElbd$j4oU^0QRtyu3*?
z&8mfI4{p@2ZwTQr7LSPpoc35*<6TGv%=*Gu@Tge!wcs_TdNiUFe^3^ljCW^bSmi$n
zD%H%#7NZ%=<K%pd_lu+r7Td5bWh0pnmt$><+7PgH1~pXO5<Ri;hRiM*VbDD*3TY>n
zv2G<zjP)m*5i?;^aEp`wW%g!SFH$S3MJC&>iR)VlT7%8$(d0M`l$A$~*$hf`zJRgQ
zd~w@Y7Z@yJKew9j2{jo|Z@I&KrHZ)U=>$?vn@@cMyvUg}a9B=>rmOW)#<7cB#_$K<
z9M{gA4?khG)kc^*av3y2_3&p}0yfxBBTFkEuC-X)YlMu)>dqNl$|o9|KC2+8>@9})
zn!{%A(30=F`|-OoXJJFdYqWLOuv+^YY;MXn{-J&_MI4N!_jP9U&ORS@%7Ofj*e@|^
z7;#JOOk}=wW|wljV0lAMwEsmh`(|%TFxv;o)RB<6o`l6>Y~S2;otnjcUV7kra@#E@
z^xij>jQ$(L``k^b3<qfY-XbU_7m7L2tyoAge3-?4fy1}p5~phpq8asR(An}(ymjCr
zm!7eZMbryL^Ulm6yG#!V<AzgrxiP8P-{f)aiNwU3=h@zN+|PeP5gWhqIXV4#em{3M
z(Jl#hEE1vQM^qyA8}~>I!N9tcbbg=&8uR^-pm&Lk4t!u$Cc$vOJcp}=T}0G@fd=UZ
ze()J`jZFJ?0&6{zK&x{{vP}&|h~Ybsb*P0;95ESPs_ocnGg|DQ--V9lJ)-sQ0(xq)
zpL*v`r7mHMAYsv<(n4!_ByQVHO@E>gJgtJV<|l{(#oSuu&pa5K*aF54h}t0bNa|=K
zY?fT<mY1zqxmTU7$ZcoiZ~kRfYsS(0RiClbrWXr-jD_R(k2Iz<h(>Jiqrr{OFhF7n
zT3kmXNX+XRyg1FT<@OUkw6delSKH`t(|8z4ug7z-r(na83G^#`3Z-t+#q2q4*qPc-
zSobu!B*|(nt&Ua}nP+>!_N*@6Z27~V(PdcPv*j;0n&8!hc^Le1F>8`(WlTPpLd(Qn
zgP}Jt#OW%Xy%xt#srM6(z4r}%$GxDtOP3{V+=NRSS21mdh&0uW;q>4edl_g#GX1k@
z(eN;|X+5Id&0btp>@S#C6GIZeKN{5Ca73)n5He4^N`Cu_MJiLpT8G|Y{A=|=SzI`T
zcNO@!D{i>rFT&fygDBwr1lm@gEZ$#J1M7mJl(cpY_n&x?>=kb?x72nV&U^^90#zCp
zoGpkuybjavmg8xpH0c;c^8O>5NomMW{@*-~_ge~>fp-8-zl)<|O1GdjxfY|(y6|uM
zxA1<9m#BJlIo3QMhs@qU92upAzC-;XweUQTvAjoqt2)KLlN8LUEoAH0oMR6XV<9m^
zSzy?&t60bPKirK}p}3eQ?2N4=U0wG~uwLpnhCOyA>ex-@nc~IsP$=|pg<ZEwN20qI
ze-Ro+ZefN9)C{2MGvshRXe|kh6-j2lITYod@a(Gj=&nA6;FG7ZNU4LHiM4@MbChVy
zrwDqxD+*~1Inb;=!*~0(<BF|2quWXB=AIBtGJGY<fBca@s?^3|n;eYxvSi(rDl||g
z6?&bL!cC5kv18Cs_V83Q6~2sP-wpbeW(>b1_E)sB@wb9SH;>Af>U`TTuxuX=z2ac1
zd^U!R{z(X*O^su}Lay`U`Fr_$lVzfJe?6Jp<TqUL>NXx_wUH8fhTw4XWLQRBqLmXz
z;`y9a=v#adwN(?~HMCLS^ZqUFjP?_J@Tn!ex&$<arQp`7`H(G(V-L0#6T8uh;Ts&G
zT~k*w!=Z&b*Ggc-s^Q4U{K@?`4#N_yRWQ`KSYlq6$kM*eLf~j`s4OeN;fP$i`rjGU
zJdPsMylvQ?B}u;iV+50wyjjQ2U;LGXpJ;}56M}{}vTM2xEI2I>qh`b+KJz?m2386#
z>=Qsjc8ln@W&uxY&SwDuq2wgE1?}c#IFj{+D=Jxuwec1*O1dHD`lImviJt+TD`juq
z??ScA71ld>1>2~0ntNrLQ<O?H8}PLUsl&v2+P5IS&F41!jBk_#dYwc>)p?$%ze!->
zK8w~rSB0|LE@)@!QgT-$#+r3Ay^gbRdw79nbj7lVgJxslrZ3!FP{DMiO0au&0&h6<
z4vXuKU{Y_eSjQet$44B*feX!WO?rgK#m^~y>|GKX&F67-4oIqe$}app$HN@EnMYAR
zzoR^u!XE9SN1ET^)A^dlL>gez0YB=V(L)PFBheP_19Qa+xYb7rRBP^+q`5stoV5UX
z@)g+j#~wR(ou;(tKLS?ugP(deh8LeUBlWZQkkS-EYO0RRWkZ@^!i*d=nRwIF{l0Xe
zQ3oEA)(Z0T@9~^P1?=~~i`bC$5`3U9HmnOk{m~0JaI%!GoqT~-jEupr1DT>T3up7g
z+VlAHvEsdKX+DCSJ;kCuQ)2LE(lV?(_maKP@Fz1nZ8pg4Ddt?vLfnip(Z%I;e8Pr7
zT%}qIn?LKLvn7!h%({vA$9u(Gi7aXUP-c41?}|cxj^bs}l~niiDA#XFr0oZz==0e$
zK6~gXfpC2mSteW2qQ+}tZ6S-j%6*Nq&O69dHVNOO_Cf!IH*--q1mlgz=<Ae)Oi%GX
zWkqeE&nJU<NW2p7NKhd8clGRh#C$wRd(XajpFmJyCr^CvK=iQDiggY<N-Gs6K*DDl
zd40KzXR7M7>0OjaB<UqcRsYJT%=G3zwzSbB(H(d!_`)O(^cg&l@<XTR4=5kLP@+{C
zK~x<{H(SebEi@7Kdq0tv*C~;!<}|GTEK|D1>ohDH`<Tn^JQ5RLXqZ@z6}E0P%kCrT
z-fSl+H&dn;u!|aGm9TU58S1VNz>6(m$T4|DKd(N*u!Hx=v8ajJudO4mZRdELak1#@
z#}+)B6+s7%u7YJ)I@?vamJfR~1D<heG{B&PxgZQqKI^e-VK_=lOVIq%k+Ht*D4abV
z+lNLV`dA7}6Z6DN2SqWB!g*YE&1r<!xlrWDjpQlrmx_j`!_|2KwO*ghzby!2OC7T4
z<<dG*8tluO{W}Do8tq^{{1cbkV2#t$eekUQ05(hd)3J3qG{5UIRgJj_i?ALZGqE0$
zQpJ#3|CQeFxPbR^W2xoJRkC!-g5$6}giET@W~ISUeHsN_?T6?e`x|q{sgl{xD{!b7
z$ZekQqn}psG|e}ZkE^xAQf(dbvp$6zi}dJ^bSPgkyB3Emg)D4p0Qt<Up}vVSr6pSj
z;*-%=(p739x4oH2pQyoh)TN>+%7%(ARnqq5vr(I|k43)Ig+;&^<W2p|nzwACHBU`h
zkIZQNe%+0Y-X64C^RuXO!YwxQy%T>hEr^!y8VNQa5Zn92ob~UwkosA}pM6bag9i*O
zm0fWMP6pHI`}AZA_3T7*e_5=36UP&L5-Fm4E*)OshP_qE++pb^Dp$S-%ds}J%Kk0a
z_+$kihaKb_bcVmwvS!T_yIEaND!;aOJOxB{QR4E!bYrw6J%6$ju43=YEi-Qj8fF&{
z*yn(^b5jv3<`x6qtf3SuThbhM1kU<;aBKMuhigR!x2E}%sB{9VVhkvD@?JQ-Y@&>|
z8IU@Z3mw}A+?}6{Vb6Wo<b5Yd_pJjy*<}b)!uKM`=>rQsE-##Pd^{T(E&(?q2Xy!D
z5?pa4QG%w}H#1dU5ZiW;qFxTbnPu)syHZ3uHbtQCrUzS;B}7$O7F^~op=>cmQro@;
z>CJi=`*1%}ROiybuD=*EauTUcm#1w%??It&93_N2!JOvfyuojR=u_8KxNVQ4<~}Ff
zeU`%99!Nl1F_TH$Gc6uA%M){K1a!@OCg^Vq%#LiwFNd3SXva7>TNYu+>oBtMKLfiZ
zkJ0o}#BaO0!_6)i@4xr3G}&2r{AC@okKAR(ehPRpR=QMX@_Bf?cOqXwEQ|-r32H`z
z)kQpJ1y^p0${ZDhZSTZ;k(@+y_U=h4Ss6-S>cw7+m{tTeFEDUPc+I{w4}qd%8I#^A
zpbSAYW)&2$gELYw#^@s7_K&gbN6pyuHJV%M8q(%DD~W%}WRsUQh<ZK+;#=D+f&KY@
zbb0wsdarh!rj^Fh=oL=%w6#?*N?yF5AjK0$gv)61@iYn_^qp-PAth{Bv4IDBwo&J8
zRjRvt1#kLNSXS9?e(aYkmWP#+-%2ai{=SjyyUj~hj!57u1BcND{Z1UYb%VVfd7CRH
z|H11YlW|?g9Vt#ZtULbzeA~{`#7qytii08y&3ncA!joCUn?O>JxCg_JS(GH^rj5S8
zB!Tun40})qdHYdhIa-cyQQAqnw_L^b14*n_DG`exD~xsc#YUePAaq{PPK&|=sAo}J
zNzCp2Sgn<d<JlKz-qO=F%2ADb#p}WTve-9sVH6#6isydkLqwTcAHX+UqSPhh>6Xt|
zkxcywQk;8AQ2OX2lDxJcuuui=T@{qFrkQuITZ8J!Q}K080B%T&y;duZQ<3R48hf#U
zY)t(i@nb1g^{-=dN<wJZfNn_GDN;kVH2qR(6otKrr-S9c@L0o&O<C}k)IX#_S@Jh#
z%P*rFxB5ubb%VYOU%=ZZm88#9!#!>o-5T12%(#0PxjvJfI#tHA3O*r2=OUlgaRv&<
zy>Yf7AJ5i4Mz?s~b6L3+Qw+ms`pqgbtS#q-jz2_thCw(jcMjDD|1XNp!>`8ojpJ>V
zBv}a|o2*di+@C{|O|mi~<SR)+D6^6@wNs%|N+k^?jdNcOX-X4CRv}7eNXYN`{R6!^
zuX@gNjrZq$O`suL!}+qa^_aVS5wD&yj?CXi5$?%g>+&Z&pjUyAi8!6RZ}9{=VzY=1
z+9<{t`83fm{Q<&v<0QIvP#Zz9LS92}30)QVardzcaLs12M5C${NyA(Wd^5+>T8m=(
zwPFm_ZrRG@`ZST1W+8e?0hi*w<lIvNdZ90hdJ=$F_jKrWt11o33dGxnr`X*hI0h5~
z@p=0~w9lQ!J5x4e*n<!xP8)?q#{;1JrWqsJ9@4luGSqCok4NpE!d^BG!<}X8*yx<|
zWPbBHn?L*xvgAFuSJg;be{Ka$yWoXY>;IDEMkhU;HJ>cjkAs3~6a02|<D>p}_=LN_
zZEiMYR?WoDk$$Z6PZdUWMN**R50SONiuYeq$@8<H;hE$SM(r~q7w6^p+%|(7&UnGJ
zRDIawm#;7`Pl99HoY_r<3wRpYmj>xgfz@9jFOhGJnq3QE5?q4gac9Ztye}sFG^A3|
z97)KDRXoYbR(O{PY@Ey`tc?>GjHJ!Ey%!E=FPCUzMYgQiNHd8iIS!(XccZc5X?AYv
zLE(G2dj^?#UZ%pvorno9pqKAVNt{syW$7^Jnyuvp<z>8=&ut-doQhb_$#9zxO3Lgz
zhII|W*@S(_+H;E(bM#@7q%1brGLfDh@gZF+FU)<Why$NDQMHzOUaDUjg`SN?%&)2J
z$m)T_Rfp4)%R){_Z4aNfq?PZlbH%!OA^f?uE|=7gVPiI0iCo=>NuGt}LAKsm@-eUh
zHqs{$H*7Rmz;>)&@(td-2f$=smE`*Ve5~Cj@bKx{q^;4!^xKR{)v%Q|T6;jHPXXVM
z>VrrjZ)o8vo0suO7q-i1Lu1_utcwxmczG|Ft#2v2(Hf0+e%)xj*g>1VhSI~*VD{xg
z1>3Qv5oZr<;423$g}J{Y+4kv?%&1IY1Hbu0cfc7;=zJ|%vZ)de`YdDBXM4k<z6K^9
zGhnbu;GN^t(J1943OQNMMjN_eh;9Q}_gxLGg&sJ#@&;Djc|)Q7oe?p^7ZoQhDEY-8
zOm7`4Fyi*;6MdLoOgKlEwE}p%urIGsJVv)Qw9u7mNj{xdpi_7VOHynxXni;P`sOtT
z9kYXElM3W(nkY;q3f2qH@<skXnauKm__h5zzxGyy!GT58b0>kBxSnM}BQL=^H-Oc4
zJ3?G3^jb^zQPH#nO!5&+&e1!})a}g`Lub>aKIZ8A={@ZnqaeN?79e^Yq=1T=IyTq5
z9{SrqvCSPGShxNlJ@vTDosYO<SN0a_p4>ofi;MBHZmVeHRxf(w7)a4$40y(rR=Qt%
zmr+%TklhgGUT=TX0IT!t<iSK?pSKh><)@k53{8G+<0|&pC?fYQhQhx!g68%<gX80G
zQP=Sr10S7s?wufmCFeM4PIZUhwg8+~j-!(gkJDO}u|m&Of!5(?L~^}$;FS+!2cH-4
zLtez%c058n+;baeou=Id@2RhsLf-hM0$3c<W3z4dz&LmvPhKeO=FhLe>^`qB!$Dng
zU~(=Vgu8OP+y#=yhsx<_YZprt8!&_O9THu~Z1@ESqH|#~4SPI~x|N?ZWykx}X4@&!
z@$X=_5;z*BOXsEYWTeg>%@#X_Qc31E4C%Or^x#CwA2bqSwL@WDRLD<H*vCs9?h=`|
zP}C<!gS0L~{#v$;Wmovpi?Mgv;j?|kfBGnjOM6|-@%lKAzCZHF30PIa6gtu%@pPg$
zS8rohqdDLFTvqtjc#K)U?yw@aM!qdx5p9BhLhXe*KWpcU564{4l^oCS`^t)6tQ2x=
z1r?Hhnfu@y<G?!&7t!F??_ivA2?aa*({=46JpVEkMU~S?w(%w|&T~iWJ7@ZAFNhFM
zMe_UmC!t*EKb+?KP<)aRMB6T7b;2zS`0R)Gmz}t^`Uezs-sB7Cx504GGiX02vb=Mf
zMR)v@tZ=PGLv|U}Tq)t|4+hZG@99X~v_y3B=24!0PciR{>2vNoyF~O^$qg4&QgL(R
zTU5T^h0nIGRBUk>vLVjMnDYt=<^3?nV=o`8c^;1UeleNq2)bc;oooLXMUEp2xaoEo
z@jQV$+mhjk_LdfE%I37Ty$_<4gL&q?)#%*%1ly!v@a6qa@x)CbNP6PU57R`Nvd#|U
zluK~zY%gIK?uw-LI0`MELY}e#kTU9q)O_G_*FyFzehK1M2s_iup=hj5gU@(@6Vp0J
zJ@=zv6k3fR)vpB3WEsUhO6K7My`ev9FwExX@Tk{TbVdCVYW9q$W&93%p5=&ZC3=|i
z;TDY<n?QZtLrHQspLe@mr~5-XF!RbnA-}RvbZPi<`f=p~oE#qJ{F!%}-3jeMN%|~6
zVLG4D`v%@h+v7=JRl22F%I1g0apU$p=3OWFTOPWzLHBAfExQE07Y?P<PtUNQEzXpc
zl9Q`(`MJT^F>7dWOc^`jk^{vd51=r<FTE%VV)ri(<p<7A$6X<(qxnOJg1-OdDsh6V
z<$feRjWD5M+cr`9;C)cgEv3-8)g*W55IXz5z^(BY*!aY+Siil6WQ%g)yXGY=*y_#-
z^DnZo(temFx=i-REZCo~=PBo&8b;Kt;o&P%u{PiU54Ru3*BP9rz_7P$mwq?vH9|>j
zp#P5-?V8A)U+v+Y@@ufFU>C+Kew8dM@?uXOC$o49X&!fMB96r+vB0-~B@XuCR6KVp
z|Do;-6Wfw(|C3jt`oLM>%^M^^1-FI%OA1AIx|p1f9sLM@gWW=ASoN4TTlaxWq;@{T
zfT`A)btg=;O@mW)R3;v$<ne&r+GyXRjOdOF5*vXLn{kk%XXY%P;cUhJTat^n)6bFA
zmO2c4Qo-BC_7*R}Es@NrDo(YDbZUDomzyo3VY2(tZTF43)*r!WzM9;Y<gr78*THRt
zH(tDmWkrI=;)C>i9`$}S6?H0+LhdQ}KlP@FpgvfkJQ+Ldt6BS+OGxwx5*^5xhNLB*
z+2^tww7Ky%wk144_iks2e%EHmNMA(m0~;2>20-dYJe<$mBf58$ZE!xvj9Z6MM%fvx
z9deDL6BG~~G!<%|aco?cGOc@9$Q7Eeu%smnKcwFwG_;w%e04^!x*ysmG_a9Py0p;H
z1Iu4Mh3!ZQlb1OO3z-f{L-ie0l~v$;oit{b6~X4$H%N^Ag}qKZi%;sstdnYB6)=R|
zdQ&d?GwC(g+<A>j67m7v+31=VL9$!c(UXhcsdKe#-kfW>Jo-c`y48mv&T|ahtEOV}
z`psxcK7ukOYi4!WoUSMxrYx5Pi7a!b)Ui2GkBa1ft32s$^UT~^myF5g%LlSAJ&s`?
z=Svp8(SzoRZ&=p*4G!eLz|;6RoLL-ClJO;&U)YJ^4{ouSpRy^S<tXpmvrsf3>Oc4`
zItzoT4>4bUMoz^3htw6_OFU2%%wOB@<J_bL)5?RmNM<=~g^XK9<aKn++zHj%Y|QqT
z=R21i!1y~k81v_sfl05=x$Hu=up>AFqu1RmF(rh$dIiHb;{tuN15*ya4Mq3!7^Hb0
z-<;n=ZQmSp-x-I@MITs<(+RwuCt^`LPF$ru8RHM-AdZx2XLtwvJBQKawO-srEJs^z
zE9QCex%5>lhu3T8;a6J#?nMnEHzx!99i2e8lFKDIy7O3pstg(HI`SPerRmB%G2O6F
zMd9D8bpNS8Ce@U}?_L78T>J;|Hv%APh$m^iI3}}s3TorpS+(0XA$Oh%^TFpKdU=w<
zcORpol24-Cy!W(mz$GmEqfA!X26(sr1ik#EL~3ffqBHmH;c9mk-M>9at)BR(aW?E`
zaT3p$9>98Q_=>(4`eR>gIVJ8N$)5bDOTP!arYDx8p?vcbw3L$Yx_3OZ0wy9i<vRJk
zE#^Lwv!s3h3Z%Ac%)M;7i4vop3$<zv6yk<JdBa`Y)|$-B+^;|(@D@4!?2$|w_#D&y
zR|-y{_mX3~<WVzKnLb6hG3^9PxZWE|=kGT$H&t1rX!e4|OlfgL=}}f?Dks)6SAo{J
zEK=F(OHZ$)BD>WFDvyPH+Q?`mWq9FTO%$z|teSUidk+0^yAQl+lC1WAOCyI&<ks_c
zvRm8x)3C63jQQ|M6sp(@vK#u-k|`HNB{Aa>S0R=6>B3{SR^<Vv<P?#dkS{fJO2YTN
z%P{V65^cDt$b-$xNv{7`(U#8w^09H3z|^JVO36*a;~B*CCc{W;4PV$#mUJAFsA%JC
z7GH6TW=3Dhy}NvfkTDEIeoGUJZqgSx%4hTwKUwxMOL{5Pd^aoXS$`hRl^*oLlO6rU
zZ%ga2j)D>WPln=ued0fx`!ThBr5HJH1T$&!rF*6|Xn5Bjl7I`y_<kGLKAq=N#V+i+
zMF{^}<VPkGcR*R@28C}fL4EER%uslM`_DY^Vd^ck7Tse#`C=-%ZG$&z3&^*&i(cyr
zxy}(e24)|O*$ksFK5eFvkP|Hs-b)Wz?t?U%^6@?P8+XzZg^Plt#|qw|8X~EZNTlEf
z`|zZThjyJ~uHmN0HCjb2!klM#y#l2j8OP?QA287IdBWDcjX-v8I1iC`pw<7*;iID!
zTX*#!O8y+B{T(OK>G})H=f$ukicXTfbLUWy^Bx*zau?g$mQkNFS@yTDG{nDqi%)Y~
zuK8aWt(lyO$w><2aC4{RbKeXDE$?K6oQh-BD!wpw&7ebj6fnGT0)0{X%Cv^9hGXAC
z!N>9h3uj-!&xN<?oKhgwW`7fPnW~`ge}&NAY$LfoxDQ<K_(NrXHEjFpVeZ&uZf3a#
z4Vg-!9Zf@Mh`>+92t7$x*)7rHFB1ANaw;hceTh!|2IyK8hyu5T@gu%I%%;y4e)44p
zXXj4Ps*J<zd%cZ>Z1u4;xHrGEKNvZtYW$#89^%$Yi|1&huzv3uiYsHO#!H(mpLUMi
zPWZw@{S^Wv4zy-|i)fd*BTvIc_#+ipN0hN3TXlBSBZ3{G7K%=G#H>G)(O%Ha8gC?0
zsQ&^!1i={ht+)8AdLxc5FQ&_nJlMf$q0~7s8&CD~x&7Y<?Co-6R`7i=Yj2%Sv1w7r
zXstuBeH@yc<!~j)4hq{ZFzu{DjQ^1iZt)6+S}UQUVNMg`N^xzW1j`5Y%2T}Bky|*m
z8cPhkBo=YESohf?nEo-N`<rjW!8b&hUq~oo$y+>+Eysgi)#!53#5~1;e8QwO9Ndza
z`}M;iHb-F%HpY0u<m*qeQI-+!dO4Bl|5ZoG?@E@eB;>z8o~NdgYWSrs5Jhde&Bv^E
zqigR3CVYG-O}!_UtkXMA9>wYSE7TvIQe!0HAx|-Kb_LSaLZJMCAoFG;Jo^So_5|cG
zt+lPNy0e`bF1Sdd6UV73<PFMtK4Yl)3H-R;L4BiSvCKFLg_iD=Rylx@Vr|&o-oN-$
z$Nk(W(VA3tMN@0wKyoUI$6&=Usy|l%;k69e%w2SGWhDN4xCyaJf_uYn2X-d3KwsxG
z3tOJbqK6wpW41lNwfio!+TlYhZe~O4Y@BHS<B_n|xN9(R>lxh3T!pO#typ9wrenT`
z*>01)^y1`bHaj;O!@}b50B5i<{h;6~jFnht-Qj(IMUz6=ALcqHoD6@+<;@E2gOx7U
zwDF)OKOQxS8qX$)#%Y(cBa2n2xT3dsW&S7{^<axg-cyzAcHW>h|0#&KztG3>X|pL*
zE)>R>W0}doOLTG~VO@m0*xApS9Lf&zpc}WSCHOP?tKLJ(yt8!sY!fD*_Q9oF@$hum
zCutj|%fmly$DoqcB2(3J^a^WZBL@bsimxv*Ma6}xQ~F32YWF2e{~Sr};6K7lCk*#g
zgK(<Vi;C|h!?^7-TFf0uS>Zc&e16Gl-pu87!SnF4-%@(BGhZ_4$!w(2CVEox2P2yN
z>7u?NFMYP0*~SU8*eyRPS0Nqo%a_9X>UF$6R?J*32=0iRn?z=@PNEMNBB|%;11>e$
zL9{m`j*YD6)Vak4GcvA2dEgCXdhTP!KSomjoH~h~aUZ@gd=H)6{|M<nSKvxs6yp8~
z>}Goo?9OOGSLO^a?(Wa6E<_+;))4wyv7TvKrc$G5DOn#|g7{sw67TOV0@Er(M{d+%
zwYM7E!P?1su`N~YEv4%PV^Q1W0_i$iCe~NN0;|JxwCf8+rZlsNuiL3n{wa+x?~T<z
zITvmCfWg16QSY>Jk%H4c?0$BRzQ%{c?RYjV&+wzU0(WhYc9ITL7)5;#M(0UGlBm9B
zldN{IJ<Tt;=e?D<UHP4LnR{{Z&?7=!>qqlKzcaVm7WBMnVM7;`AWvo|Q#X~OMClv+
zUU4FQzTPf!isd=OhBhI`ZaDk#w-@AF2jb<RTVRj+k%7lku3qzoTZLcY1I@Qm=H!{Q
zB<(PUMir2ZnZT<>jbNSk;<(x@nLN{Dnmo2W5?-tI$Ys+*W^&4(q~b<F_`vMROU?`p
zOewOin>x!+bIa6boCwOHro3vNGB*KRUd3Z_<0fvtdm|hwevrHEQ@T?>4_o`?VXBTZ
znRGnomK&{U%pqTVw3<%$MmfXs>tmSYoW*GycN7=dQcnLIcJ6sRiAG9j)fjoUTiYKQ
z;gjf(_XbSca|gTDTVY+*Lj1Z@M1vbHqxw}*u4DUNW@Q?Nqsrgeb6YnkrKYnb$rqrY
z?g00G2bpNtU1ZL4h2g4e6uz<ox2*Hn)ma6Y=Aw^sy+l6tjR#sEeuL$LR5svoJ%04Q
z4$r}lFy^^GmYLn6B?)Jtv*9cj2Z!OntJ}z`f6Xh*JNWHiFR3!y5E&a@vpEj$FloFF
zHVD5*{Jt)B!ZZl?&ie2Nap_F{#Sr}T>x1Dj^>pr35ig&r!?q68U^1U8(Z0Kg?)=M@
zEYX-t%BG$e`YML9cKXoS`y-jLu-nvL>4&=)i+N(<2g%_Hy69MxP1#0$gq~nE4gIu}
zPU{WD=av~vH_%sLN4<IN=WV1rcQ>V3YI3a^Pcg&32)5b5$nZKqztr97{k5z7K#r8S
zcGPj<y)hSCHuc4l%I`cPK=_=&`y_SEy@XtvAw4`k4x4``P_KTr*mc^QdMsP;THz}`
z+L)7ruOsX?rt&F2%gJKGCyB$>bNJ6}9H+o)OlfprWa)wJE#a7T*oe9sl+jha0FRPy
z)A;A+=)HFq3dKhFZ&4$ArtKr_ncPI7D$$T^5PTp_TSzX?i|Uj5kcO~RP5YDu^U^3R
z94zehlG~Z>{0vMAYDCVgt0FV+#S*8<C+8^#CLm^<DeeWFgxa9pNY?FQ@vV;~SI*VZ
zG%W@4uo%aVuW%Aw7TlZ0rt`=sw~oz;8jM#J*O*ISDJ>Y}N;(SyFnj)avWUI`_p9w_
zlg}2l-Omv<Yk9C)iP9qfOu_j%y^*|ThOnB5a5jCE0g?>1iA?uMi5tG$6Q2A2VC$_Z
zTHs6$7A!)<@QD<3Ws)#oa3Z^^;iR@{2KBS$5G`UfXmW3STltsGmAQduCRc>|G#<N#
zMp0h13X=ZnP}39{aemnhY>l~vj{cRnH|K)jWsS$&`TjVN?M!z66;qE-Bug>#6%A0X
z!4<xRWvmY2S9X8KlUw<)So|!v<(Ur!59mVN(UW9R>`Z^Rgww4qUHbQ`92+mm@e>0E
z!8campvCz}@yy|IWBZ6JH}_CZ_)g>tEXm|e!|-yMH%9EaLE67l$#7aW#n)aku$;h<
zk-e6yj_{*x3DfxrPZ?ZWA0zR27Y=FlD7rqNPPA%fE-P*^=XX|>bDLjQbhyoxmX$2z
zo6X)Ey#AiS6?#r0<XwN7ny>;lRg0mknGB^fcUif$94e&*hM>Tgea`BGJyn14@=hp5
z2|1I*k1f2|;VM=prf?JEY4EI47R&UG#kRZ#c)gPq&)?d>8-lk>4mEV}<RN0{U%QEV
zpNss*Mn!zsr~|8|1IUZ}QK3=`)N10<Th*5|9?Yj=S!+Ihz9Tg^=5yWMd7{g;7Wk&*
zPi(a%H+A#F)XQthB|QR1_VyMB*+;SZtYY*GS%IaqQn6-=@GP0FM5)?&x_`OM;8%|l
zWmkKVOUgE!{4yRdemNt0<u@EYoq?xwSF?!8mry%91cweu^TNHmU{p68&Wj2W(vnFH
zBj(Tuy8wLsGY|GVy)bA%8NIGdp<rRYW_|tui&x)33k{awXQU?L1LY~|NehfN>d^X~
z3i!|CAQay($4=WV7_qs4O=^kf>&()SYZ*vhe;N_GUK1A6HyJpM_M>5bQ)yb(50>@i
zE(RPuk4Go2)8@ELCLJ701CF174(4O_tad)#2hgfaW_iC=Sn;nAY=83>7EkZ;VZ#S-
z2gN`<YL!9w++}>-I|uCio5MbCdV}TH{rT74mT;4N6D_$bgB;akl<L(e5wHG^!J9iU
z{?|Q>sa%{pQ{_4HoSRMIeHO7P-<{yD^#Z@rJz%EsfTx)Cr5j2Map=VWrnSI?HipR3
z^GOOU;awg4O49hS!VLodcNPt@)1VPqf{qk-C=GeWN52Z>A;)59+w+HP$?-~*erm_6
z*VoC~M4iG5!fEk`DyBZ!jML)T{Eb6CJ=oo3aJ03RejXFk4D)<m*X+iAy$eHf<XJeq
zY=U>vX?Ue=L19}AMi^?6i;W!}I=clrsdc#Rmw^9dc91yl2JHKsVShedmaKG>5-(mU
zp<hYCr2l6Q!Yw-4P76;G-`|c;Bg?6>G!!9gUo(@63n*_JC+ZBmhr<s?qYanX*YP6i
zo*#e)jc+t`QzkZ#=>uO&;zn8i@LN*I)XKfkr{)CNg<eDF+<kn^?=WmMS;ziJy~Ed~
zC2;ah!R5K7{KL^_Y|5`HQQ2LBi(wW`ydA^}3r^szr#BiNaXw&5kZ3?+5|fzEK!^MQ
zQrscz{iN2=nm{jJxZx$Tj~P*)V0%>kxk~Cn?E2cWyRd&4C~y)f<X7{Q?y4t{LG)xS
zx2>bKC9V`uC!e?PS|C!x@<kL-i`UO;k$UAKzq_r2H-AXO?6k9xUmAm(H-7Mm2QI<p
zeg$7DDq;WTj$)s;U&d&wWQ>+2)JJ}TX5nDWO6cTws;00IJ8PQmw1i%bUJdnmw(wT|
zN)_ks@|jw$)OUXqb%-)4VRAF+jov_~j_kmbO)J=jahi}_S-?b##|SfOcXBiRgH-wF
zxS}KEe#0#wXQxeuEm3^rWqalpAT6rWj7GD-tXcNn$0i7T>(DZlymcWVtnr)h3@%(O
z<TzE~xGN4Tq-Rr(P+uEn8_-hCZVG&To1UJT%U2IrB~ffvqpVGv_@PcCzOh4z68-=3
z3s<rwQaz&FLAoLAkxVFT2mV8h>1<lJT8SQ+*FtVh0*nOzfvD7#PrT|zms2}g<rF1x
zr*;mXv27)#9O(mVsgGDZwI7-43GM;8CiE_NOjd$7IJ4)Vz^9iGyW&qb%D3R$;D`8Q
zlu7v~)=*X-J-Am&4Ei5#L!67EWZa>}>}RqMJ>B??ZQEap6DMwCoXI1eJ<E^`Q{FPW
zg%PMb9FH9157g5yhi@~o<ug7%6Y7RPu(&jUE5FjBSlNrLZR<*?yqb;7<FS1HT(4XO
zkDJu?!W$xo6vTO)5cR!viB>M0gspDR@h5a6GTIN*zy4<=xh56-X20|J(tTaBCgVE-
zBfi6C={2m1ldxyfk8yP0d77WL1F8+V7|~HihEYz?P}_s4&kF@-w<F#;y@cDNLHwJ(
zH@XDAGUl)+b`MFR=);-3{^v^iAmo)=H*Z6LUn6W5>63*@De4;CSaS0`ihtn)$)VXa
zZ_fvMVAR5n|1HIZzWr$PMQO3q$yy!|7R%Mw2T}EjTBsOHm}Z+6>22T3hdlVkMS25q
z;Kdn!;$Lrxib*cj$bIBHPY9m1%pxeK*wG#Ba74d%gr9aP>HW*c&O}F%VSB8kc~UVx
zG!zQ;i6Z{H?!_Y42a<#16RMX_ps;86*t_T?G7>y$zO~gb{JIBC0l}<(?;lB2NGy`W
zLwU)6k6F7=M{M}dmZm?RkUMSe4EiYeA-1JI#nGL|=;IJ;F8AUIDh+yzJH3;rYI`C}
z7XB97kGJsE=2^JZb(JL;y~FqOYsflhE$Z|9@w`W$>8%|H&-xTh@0^QuVne|ToyI@T
zlqdU&v2;JJicN@A$3~TO(SHvovB9gvh}iQOc6IjbKih#Y8L<(;0!OX(Xo{%fxG$o6
zT$xo``MfPc-4J_fH!6C!(VkuobiMg8FPa=d&l4+z9M@<>FI>elO&8-@@+g*EdK)YM
z22!q+;7k%s#Q~4rNzNSGfvrPELBn1HDk>L6Bfnaa-I#v{KQ0Qp{L)#>*yJ2_H=W1n
zf*{=g;m8Elok&f|3NLp_i$DBUz|{8M;sGU7DL|<eC4B_v=LVtY374na?_=oDi5N^#
zTT6@DAEWQ#P*HBtZ*ut=1EaSt>~@MCiEGl>#o0R8ek~H~`lnIY@hrIJnQ;%JgJeTq
zG~W9t42RZ>QX413yKJ3Ap?y124R*t65@E*D!Jv~4q8nz{B}aRVg`UZt6)x1rq5M01
z_|aYXI_V_(ubRV7d~idu(PKJo8X+2&QH(iKg;>xsgVq<-v0u_6JpWk6eP=u%J+D59
z_KBcFQ{vzl=1v(sg%sqg3)S!axJd9NZgjjT<l|3^hA}<l`%OWt$^%~ds-0pL2jWPL
zTpnw9#5T@}LD>8l=y%>F)jiE@rNvt|H0L@@i=)vZU5iydsq~}DoTC1=3wyhFu-bbG
z`k5{`cIX+s)|V#BnfK8=$%VbDRe`QNqba=vx8s<p@EPGtrLGp-yDEd3$EIP;<@K!L
z&mKsRe#Ayu9iqY?j^uP(LY>K`Bo(BFaK9hC)<BrAdpq-OU#^0UYh|~>T<N)TJx$+w
zgI8aeh!4TnFl?KQXzR^%(Lw8a%Kq;h4koYQcZ1)-^R7F>^9QmM&O!8RN(EWGBqX19
zr^SCYVY=FbvSMP9q>zd-_7GpjI^d+N8nRN%cwdhQD0{!A!U8{tMtn#5zZgs$Bu5$c
z_t?>6?{V^Du;}|XEw*P@8CF>K;UjZp#l9(~_@?*;tv;{WQNc5pOkx<C3SO@5li{%P
z7V64t+2XN1lB_GrJm*6#)x;}_o8=bKn;IqYli_`tpRvBk+E^y9ZBL;j-o66877b-V
z{m;VQV;FU(dBf?%P43gVk{$bLLW+9vI4(~#!MzM$KL-lTzXuP#qmLh9@`Ce1!j>nc
z(XgH2RBK)ckLnc^os$To*k;6T&cylilhC}QFLq7W;NA0+Fz|T;H!gjQy3tqZZJItN
zt47j8$19}ppc!VrH2ARPX-HepilR?JOr!G&<mGhv>w6|l$2XDW1GM1h?}6959*P$I
zcEp2^ckweoVAuPOhw+KaLT?s=+nQZW#q%IUy2J;4zsFsKccke!IlfI~M#m59FvW?F
zCC?JB!)n_&j1DQrimWop<CPOo{$nr72O6TL_5UnQE!^JB7j2j?OIs5aXl3^tB)>k%
zX6W}Ly;aVFzfB)43%#gE;4zXi1y_fiinw@<;1>~`?=8EW*^tl!m}gu}kN3^O1@|i8
zk1+<xPe7)tDb5{R%jd4l7ye%pSzWJ70z-EYl^?{28?TKe!PV@I@_E+qHc2$Fx{S8_
z<gj>^ak&a?7Ps%`33=sk-sEUNUN_=c_no1rI=6<3g931&stjfa2)DwW@kLX(1#&zT
z?I#n_<CP7k65pKQp@bEhM%*Fr7dy39L~es7kj4UIVtv-KR&~KAejouw=O!R^goEHR
zD_|A>g}^j_EU)}=nq1XRv7lIA<WAT@@k51v<Ks?J&7MqhQWbpMkDu7=aZ6;jBo@z(
zCt=C68|+QcXej6h(Itf%8kimg<0=obz9D4J>tyo^J3_EzX^6ym@J$l=4x>?b6va8y
zccAC`HEf!z#eOfniP(HiN#CU5v}@Bi*#F8z$jfJf8zdHXeJ`+4=Qv-tTAeOsTqawK
zN_2mIjko9TLi=Yl3Z|Z<#2-@x-^nhva`ht~{#*)Gim#w68bD({-Nq41sXW(jGUC7+
z2k7aGjj){qNY6573IUI4Lv=cDd=?7z=zK}L%})Ht55f4l*`)m~9mTKOd7YLH^|aKo
zT3<(sQVFA$ct;vcs<=2upQJ~c@Rf@vp<~%3k@VqvB1ynN9;tW>ktUohIxlj+o0VAc
zT3+;`NKJC$VmeeibXlK&5?(*^jp)!mIc8>mP;mKvlYpv`nb%9Cr!f{QgL1gbU}>?<
z03Ep71(4ZhRWt;>6D?JFLggo-V6tcq^3~&D<o%d+J+lz$EuRm?z)9RBszzWy1z*}k
z34QPNhuzP7BH3mT#YSGp$eq>nj|zr{p+r#$YTtW_pZ)D*@!Ml5e{~@w4+3y`N(f(B
z<A4hPB-~f<qshk(kag8#x^3CWn@vxmYmpS$L<k;usrRr5Ps7MhCu!_TF8HQ{VA{1G
zAy?jT^+QQK{`Yz)K77U|r@SPMNlJOuWB1{qtR*QWRinJGHO_lopu((ODBIypih>8%
z@r>Yw->f2D8JP}eTY(eK(4@MyV|bZXV-QoQik&^9(XUBWBI-B?y+A1*H?4!Wp8bu)
zGZQI)g%_gl&7p;NopHpai0gkAJk@8f(3+0#>}x^_3O{<oW|9l`eGa21{WoKe_&w#O
zEvA|~z4G!if9GtrdBWWn2>nWR0lq&pzzMBt^7;1(X8oLDR`-VQ%2VZXZ<f>6C{MEg
zw~0C|++p0{OpYC&M2qDN$aPtf<n$CyO*2C=s4iW|wS8kEHG6EHbe_33U*~`Fq{YWC
zCP;d$|3Ob+MB^-7p)f3!DJPi1FVRb)HbBU2Jx$}vN(~e#yPdR-HnZOgJ#o5a1-;v_
z6f4(wNH*zw;Qy`aCG6_Fx$Er1WM$q@B$t&a$()yuAe#bgis>UZy!jTRA0J`gpA~b3
z7ZK>p6mn%ha>&0+8DD*>VA#}~u3jY=HqNGPZ&U>?zg)<y3z?`jVjl9}Mq!p!%JfC4
z^gK2L_FElMW#Phd;~TKFdb&i$_$$A*)S70QaduMEo6X!4fM*)HICLTp{_mDjVv;T{
zovdV(;eZ93exR>iF~7O*67|;cpkpo1Xxu(~I(AAKe3ekUnaAKsOCr6s(h!{)P)Yg6
zC-Q=bi8Nu785Ej5sQOR{UdlX2<i8($x?f}NQ9csRyY^wu=}i>ZtO`YGPuiC5fcUW=
zax?xjV2?iTM3lG=-2%sCxvrTUrJvDOrwL>;;hjXLAzXC(b|+tR+?#GMHmCjnZnIIF
z^RYQL8g>zhIMnt5np;fCctfV>uX#H4+*0O$Y%)Z0Y!TaYN2p;X(&B2lW9a>Njp*a<
zflSUanN7Q?Nkc22vKZavcs|r0S)sQPzQ_af-#4;~aTV;vfnL!3t0eyJRzta-{_LPg
zV3^jJ!Cs_|r>3sBAW5dDyW+^|?Jx`(8AaXUV|aPY8<G-s@&U2WDWyvn*K_wm<H%%o
zIHZZCZBKxYcR0M~CDDX?wRm15aNOLM_Af{zxo(d9w0bsThz^-)o<qWq0xq5w3zy~z
z*kihY68l`oB0nD#X{qDa=~R?VDkIk|nNYm5n%o8UUTtPRiVtofH(51aaFb!ppbi}U
zByfrK0o47&M#!lbQn$`%3Qs1kkdla|0j8vR#*HRih++Y!Vi3PyTKwy`JoVq_fh*@T
zCB?lWFd;1(L)aaRnlluejfm$MHj`8N(|KpjULvJF6H)J%AuBcy@8*b*(-nc8)}CZ7
z-@>)Gv@z#>8fcBWOmYLaLH+DG`iIMulXn|>TT5_iwy@hb6G5*!gk3ciJWW$(<p${#
ziPl_G$cq{Mk^f7%2HTw$e2@4N9%!0Vva<ypUVRZ=D{}<jashu>H;|8YAHe8|8*DbT
zP;8bSEBg8qmoI*l9C6TMcmMuJ(TievzOu5||3fBCKN2l^eY2ZMjTlU?Zr-OX_Y;)&
z&ygG=NAQ`u8ku#(TC9`XKr`}vXy+OkZs>ZRrY~JhQb*(FrTX8%x|=sCV%!~czCS?0
zv%~1a$Fnrs<ThH~$I_f<tp>M*I~IyX&h!Kax!TP&qRzBtd>+*V);^uoj2gIZj~$&<
ztAp*1hfvpQVD9^3a5pjy4o9ot5~+`{ix%*9wZsPfRv4~1Kof7Jivl-X!mka>(70$I
zB`aK{_lj2#*AdNA987rJl#9aL;52QW(8C7g+L1{6u&78^h4M|j1*Ucyo<zmby236}
zxOo=aSv@V+vxCi#wb<{t5$==oS-%e#u;pB@ywvhr2J-(Za7k-B<+|q5xDCLo!`Jxl
z-HLn}#PsUN56pABM@sut=*RSaEUKs)22XF0JGl$~I^mAh*q`iA?MaF$-pQQiG~k`g
zHqlV!RvbJ!lY56n5#{;f&gy2NM^dKh))BCExXO*b3HL-QrjqHVH>`GBC~OASNftEU
z=i@wu4C<%Lka{x&&vh<i@z?&in5vCW{_1qbC=xp6+u&Z*SGX%;La*Bt#s0rm(4R$R
zNYsqT`MOt=Ec`=6|GIVQ#hzpYD+(M^@mMC2Goq5IUudAi1XB7Uu%ChJptfTM<>ySp
zuYmEWD5+rW#g;hA!pSXR2>wXSh5I0G7^bzA+U|}dy}DK8zhEA`g<bORBTHy}VmhB-
zI$1Jy>`@+gF_*^fFG5%MO_bN~C(Abr=-g5xx~#sHZiKdSw_QgmKX4;frF^B6yG~)|
zusG=Tejw@@I925NR~p+_2+ZbIPaN&1Ky5>+*gC_l&|ChSIT_c(>As;s+@es5wi3Jm
zMVjO}I2(f;8b~ehJev27WtmUUk;UbS2wHK0#hZQ*_>BIPoTNqWY968=372S+l0Ids
zY{h@_kvMiZi-tA4%q^5ZNUnv05O(<zhSg-jrtcPv`m~nKt-Qz|NY!x1Qy%Q%#$y;z
zl`7H;$;K?-D6Fp!M8F6kSEA;{U)&&an$%67l?qs@ZzWS*<_PZVht*YjQ12lYr1*?)
zP`@R3U)n|a*?ZCFvz$0?_9v09^GVcPNQ3si$#j2w1wG82O2=+w(YEz(`9;@GX8*;K
zmA!T6Me)mcil!NKULQk5!$&B!s6cC4GgJk(VDwf&Js|vSw~J#*`uJChtM|s?9l{KD
zva~q;eLkyn{6xCc%mO>?*)px`s66`&$>Sejli*T(y}OOFS_S8w;YU0;GXVMSp_uW+
zg~C0DVDgkPTvs!lFY!_$dFeI$=ffaU=vjq~y-mWNO$QO}LKdXb97RQ`SWm0@`3Q5e
z(Jl}jHVmfjW{EH}TT6+WwMfn}rg_VReEzB<2(+umtCNvP3Jt-vUyX>57F_1$Jun}=
zi)~RqB;03Gf@9t$*p7Qk@3&2)cQd20c#14+UDcVQID)-9qQM^9SCen>YkEE76-2av
zlR^Q?%@}m+-PxR1l`LcCDBAaBJ;|8=!^=D_D&1HPOYw9})f|Y^4*5(fD+Xnrb~HM_
zSlDUB;g2{T=iA+QoxvxxB`d<`Tsn!9GO%3#7Z-gQMlwDI<eBDy<OVZr3tLPpvewfy
z;j<r63FhMtHKR|-1C}`|3&k3jDE+)UO%>Z?AT{$7OSAd@b%FTidmP&|l}PI0>)fQS
z-^lkpXmDZPderjcJZ(|$ygA2mu*dxZ%j_kMhDafEbtaORy!p$fC*+~JG754rX}Evj
zFfT33Gid*oive$*BXO}Wq7=rm4-ac3RqK6(JSKyvvQOTDUjfXnD**LpW>T`$G#Zjz
zFJy=|k*>lw{@;WSk%IYJv>Tm=$~0Gl`8QQ*YD#b3GXDxjNSZNcLpzN~bw`h9DkYmc
zL(%90dphbmE)EZeY>6xOOuj5?+cOJOd%ohW%}B&22hffVS2PrOh<1FvO$!`e@$mPR
z_#))yo{cs}d6GI5Wi2Sku%2J^yNkplr=Z=xoUY!U!$18D<!6=FVRNf8tbev6^=~=_
z?^lM_)8!)BZ?)7U=K(KKKk*XBNBq)^J7~D=CfQSdo?Tk|nZBsk<84<Ao8OYp{l*Bn
z9pQc8zUCg?E!~Kjji;b7#lXOBr4Iy+3X{GlE1sC;N9!lfA;X$+h)G|I>b5`Z@833h
zUH=l7?A$T3_9d@c_nrktFQXJGDO3k|Nvcl|Lcot#XkHtR8DoQ}L1Qf){8oz%CpS_1
zfBDez(u3I?dGWiM!j3oal|=3QCcHQQi;0H4#c^H24!dA5Hhw*elWYU|Im{I;PrAeH
zje3#fQUztbdMDJ(Q!vu!6*+vhlKkD{CNS^LxbX2d6^{|N3%0(@vr`#G1J`1~&p{aW
zyA+2$-J+6lirCh4le{MQpiS;C+uwYirI&3-!l@i;UE)QZHG+GuUpVcIkrsO=4-svU
z8-Zy1b7cEb1lt$Bcsj9wmwY+Sw6BJ;QOb6bBiHRHRU?l2ulAu&X`DQVG{R^3MY`fG
zjgNQ2`Jnz6nDv1&-o8E!Tl>bqDm;?>LQhfu*}L#U`58WT`H80GxzWoD*`hO(EFpVF
z$X`Cs<)!_ac}vJCNwfYl3@fR}4Wkg8Y}AC@_agG|JWU52T+nT$L7F_9$G_T*%sGm1
zFVCR1rHhc~)`G!>=P`b!;P*ZqK{A_`VWx|0p6dD%vK+dWO%1=s_P&__Rc=7b9|~Rp
zqa+sQ>o0nuau~N)3%*sYakvz{jXqx%`1jqxwBf{5rszMGDO$VYxkfZj4j&AecltE%
z%@w|OxF04QS}lq2?jv3{JdXyMT|%$k4_Qj*CU(5kg+lH`WB9Z}s0n-3<emx)U*S&8
z&S}^f_gqwc%o~bPsuXH4GPl3A3V2y0LS2kuCgirpOZh=2%3I*4C(}WjJZkxOiW<Vz
zXynJ0+~(w?T(#e&wAn75bl;vuy{rmN4347b(E}wR2K%rrZW^gqlnN}K;1Uy@6p7&-
zG;OO9PE3=@GgmF68BZposVxvv8poj?cMM_qyZPHUV~|<Ag37lRLb=Zn)-NX(eFqK3
zkYh8bEaxfHGqw|Q(?8j~m6@#8y*G`SoeGmnyCiWpf}xgsoq`KuX<J7o{`9?s%HJDl
zw$5ziZ-{3InoncK#c+**A-uukJ}>)w2H#{qa<dBuV0~`}Cj7}m()K`X7*tNJFXAyk
zCJWZ{yU{qo3@S@%NW5+eRsRy9-vOZ(`s8BZl2=SyHbkPoYY`PEM)GvAwctq^g5+lh
zh-F4`T_Y2weRmLouKZ@pbkF0Q45RFVOp>vZ(00ov=Hgw>2Ccl0pR<0#G-Cv_vySB%
z*N-u+6dwxBJjORo5WIb3{h+Zn1k)$|WnukFu}rZ{6r(;G#v9f`%IXsTY&IYDGZlam
zCH};67x(TF?t(p+6Ms+MN5_TdE&8G*bvCQu%E*3PQ_~H#bJ7uabph#0@8)xcFUG2g
zMl?jS6q==0JZ5qd?{?Y72gr`geO1)VKYVt;n=*Y;o2`#)8RuB$IwPk0y#xh<6I#mn
zC7d>=LHFWgXt|nFL)t(5X^$fR@EaKTHXrs{dy#Tx0?pZ2hgiF6DiRp&jjmcaGyMxS
z9lwFTZ8_}J{zPg|Ho*L2QsUBI$!HMY%NcooEo3JhM906I2$_9@)JJ<`klrIwlxtv2
z!-<v%J>oKt1;~5RnX6uTl{u)V@j%t_w0?*g{aB<bK5x*AXI5zo@8{nL*=fQK1m=->
z+G8AA7*1CY)(PLC+i}tVIyBebN9&7EFtJ*O+$RSSKRufg4{AuJZ>hlhf*8saHDTwJ
z_bB>ej$sZ$-cL)Fi<ZfY-JAnxNvsnh$BdxQj3H%o6z-gF#;_b$=uAzZ!k00y4x7)@
zjeb(=koy#@F6<yj)(f}MA|ca}fEhuRh}gBA5~Lo`gnyN6?AdSl_#puk;|fKcGF8yn
zBiv8STSI%@s<7|$Nb-&Fr=uacSS?pcE^%umrMq8|C7;j6-VJ8`Es7*=lbc!18evYj
zMwS99esHs1vn3IaiqLP&4XE$!VH;=8;l~XvxNPoGdiP#I6f!h|UC*mVU%hMO`Or!d
zm~a)(r|F1>jSb-+8z-@iR(G*r?_0QdB=a$mm)JUs9E{gH#j>0)Vvk=YBu`J$p^h%t
zog0?xEt<}3hJ~SX+E;x3SkB?+ioK>GSo`r6cIar3d}KZvhL2z_f>TWXh(4=3D%=)H
zkQJZuT0=*Lc}8QX5nicJAUYmHvpw8#!Kw!HRP$gm{4&K#y=2d4M)K)*_L9`~%QV?u
zlRbV@A?Y3Up8qKfk+`Mg37pnZGJTqhS8x~Z2A}7CLT;w<%t89wbAS&BSqZI#19+Ed
zEK>R<oi|Y8jC=YwVD9RKaR%C=0iOS8{2sv<Q8p9#GHuLf*lyC9I2cF6Hqy$BD7G}l
zhh4Z9gxs8we1Ew!;x68W%2X%nf9EzL9S_n)mlOmyeqsGuNV2{@2G=$}XQz2AZaUYJ
zwr3e0eUGE&taH%m89~dw9Ym|}|JzZPOtnUDpz_Zjy^dc(g?WES9Z47czM2lRDRrn8
zr5n8H$0ds`BjBRl!78-Z@_5sCZ2Nywq<u7<otz)e->OU_`+}=b=>8%q_Ldjdjy#WH
z_nmOiUFiKAZwOA2IyNus2o5%%5#9?!kho<mE|o=K`g(c%I~a*6oA>a(`)(jm5=#~f
zLg~lBWNP^`2p{e|6&$aZi9HQxBY&-6##Iq`sAEYDtM`(^IvJFGi4b!75(7{DQG%y<
zAfko+=}94b>270>F?<VM9U(2a9bH)D^!uQ-p_Fw)gre#(l(6X*JJ<G)_U&w^l>r<0
zB#TKxAM8n$=BLs7qA@N_EM?M0<p@|=%)IS%U@&nY9Stj>fg|Lp-#rfs7`O@!#{h5N
z-PG<m594(G=-t2fBHND{bmWx`hAjQccJF%v-5tM~fpAMyDYGw?+mE8OL7QM<T*{1H
zV(9rT9eldB9QA)gNT)QP%f6ornZK><(GC&y+N=#bsd84=(O>M5dW@E=SciLs-W2!u
zIzn@*=)eq3Ny`CkY^Bxkv-Ct<UJoc@A$nc8$)t4Z=Bl(_WO|C9`IMT8JY}Li)>=J)
z+tOroztM&H=@^pvBtb)?;8jo+c<Pf5<S;8rq&eP<oZ?H^`Gek&c@lyj3k5Q}IS^f|
zcTlZmB2B!02bFOTXymC3sIkp7YW^mKJlKSui&N=Or7-j95N@H?<cXGiDy4t<@0r^*
z1vWkQ8G1&yVZr?o7!zpAN38El(>;S=^K1c@-usGYD+C8nzdF8H<VDF{a<Ei!BHs5b
z*3Hjh=1+XNjbZ}pLK*%QX0l@ihN#Nj3|p7=(3mib&5qm7PP|J;<f|INS3MhQV@jFh
z{-e}gIv&Z6A_`5*fWuQaO8EGSi(0ej?&YJnZrVtZn}R5#Nw^cy&xv|Ze@sXBjzP`K
zDE#+mBUJWjVrFU}42PNE%dc3n9(#hm?H+}NF}WD8|Ajt$FeQtlg$CRGl=JTY$-;^Z
zXAB>ag$X91a1XH&1w1psDmDuvZbp*9_toS$ZWMX-JmAVAfmLi9i?-Y|qLl}GL;If@
zVw~<^(vogWy{SZcac8-yxKYSs-bKN+R&-C`6k~WDF@Il6_Wf<ayAlPO8Q_Ae6FX7V
z>`HI5+zhUS_ZD|;pAFg57tnKOIVOIsVkWoGLTc2_Y|XA1d=>H#JMV`xyA^qmk^4#Q
z%Kb39dk~*x@r+p<&XOda-NQ;=KO(DBVPxxJC+u7Ov2M}~8lE*C`cqGlrQ!vF!;z;9
zKcU96%%y$v^yz5IUF`TU1p^-m_l>=1UhwPnl5T-{a~B^H*l@uQG<_4gjs?=pN>7B9
zYE%2(JLuk<P5;WDus#CMqB&$2=KL!cZZ2_}S|-O;Os`{v&sS_V5Fypjm`&QDL5GV@
z(9P;9Oq>=5hoMd+6BL5KBQzwQc5Vos^$44yBrGB(o8Q)!XQw7!=HIU<ibGV-3jOs|
zDmY9yn(Iun9vKSvUS{)kD<jAk34ClwG6lb0#9OATvZT@m3epLrn(&>>#=bhI>)S4o
zlc^oQII|MZ1~wx+Z!CW5tJ3^M@7eApN=)B%6Laq;gTB2~x!KWns+_(N5fk<3>UUd2
zmSsq)UwhDQ<tUn8Yb2R)wu604xJgAZS4HXrZUfadY^tj&otk+UA9t3Ksbdri8rv<I
z7u}1k+U|s?=?@GXln5P<7h?Z*DdaVUqVEDvd^~uD%~Kym3T++2ZS_?cylWt=&OD=N
zscbf_L+FPlXR#Z-=VDgHOJS}!3gt^C;I*}oWgo7_s=_woT+LR>w<u#K1Fqxe+j8MP
z=Lh^6n1!3cK_ve49?9wfxf_b^A;?~X+IlUeL79b;(<20@^rRnLlye*h-^9SY<s_P}
zR6!IfLD`T@$W47H^n|Nf*99S`^?W)OP06Q^T=4SwIJ2(>U-0K;GPIkI3vQf=EN%T{
ze2BYAttZ@(SX3a~oSqNYQg7k@fN<NcLdZ=vHnF6JC}ty{$CL(0aP{vwOuQCM)`P>C
zi=`bRE0^N<nYrwtY_GgaW<KZ_V2wBKU-4S1ig!J{E7IU)DD+bj2cFhpr3NE$L3b5x
zm}$z^+>4-9!BI?apbss{nj$%V=PDUae@D|ogzQn*7q<F}ABE2^!hRCwg#H)MAnb;Z
zl^j8MF{2*S8lI#uSQOOz5Mq=fNb!pUg-j=AAA5^7H~Zjo?#JBgtKx8U@O{y(eg#Ms
zGG_%<B)I2_AugCI)NU8Z<f{#gSA@_ZJ;BN3{y&b+!yU`_4dYTqlC(8MG`08e-sfAi
z&|XTCCSM_KT4u<KjO>I|2uU*D`=SyOGAgAdl~PJcrTx2)-+$n6ywCI8_jR4;=kpGM
z*H2QL*a^Hd`HOjj6WPcjI~q9mj^sngJ=$RvgW8v6s5d*u>vv{zy~KqwlWz<8ovT5(
z`meLJ>CjB<scXd9+RoIqvkrE4FU451EHq{=rg`<HyzWSzB<j^t{=xJvl+|2e{34Ud
zBv$m(BL%|?5^-uDaaqz0R(!b#&*qAKm6nn;<eL_=NzW8n>QwrAWfcE3d4nWRyqoRj
z_zCv^8a|8mP(a-)logJn;RCBkd$kotA1t822}_ycZ3E%dh{5%(7HnS1D#`ZHQf7Cg
z7;_$F<3P}AXgsk;>u)9L^{kmFAkLlKvZ$t|lzZ&S;{#TW!kgm9c&Aj4wTi~De5amY
zFyI!<%l3krrm>@eXEErHHU$`Wf@|V5KI~CmzT4p|l)bwrrfzd%?^=%|Rc|-V=~l?c
z_p9JXYSh`3vogFe|Hn%bq?Gg6LciBWZ;YAW6-$!(;>Pm`?0%nx{ob0Gq+vt%#AbDB
zpT6+;^+;me^Ahx|R*|CSPSnS!^5yeZWBKh`y3Yr}=(34y_vYn-M{I-B+5gdG=XV(C
z;EJ_YXZWacHOg4<o6q{;1T*a?IJ?w=5{g8J+$C1HOZ(C9u!}6=OA8d=$av=Tt@MN$
z;cx$VJXB2<d%v5kyJ;3r?X5;hqt76%Qy_)tM~WVf59{?wlRTAoz{Bc_C=e9TRyjpB
zx0fr9FE?TLn)cA!m|Jv1<)q-Y8NqJ+9Sq44v(=C1NZq1__g0)u1LcqKxe9KWxUEY;
z(?bvJy6?}@{*C1tGn7eY>C0x#ej?Za8K~}c2F&{sO@E*Q%a`T(hA%ENpUW!HOIGLW
z1e?C?dpKn@i)^$5VMwS6UQAQrCjI=7<RIooMmJ@%MaM<Eq%+AL=b_U6K3Bf77>`;+
zNBL(Q;`&FB=h!Y%yS-sJHToYdDji3*{i?a+fBE_k>o4=j$2$?w<VF3uHpzOeT1l6y
zPatyJd#vtu1F;eB>9>(Rl8=ge?1TXrygZRYGjEb}Lm0No^(Kcs#lmA|1)ZX=cy{O&
z+^?A6>5@S-x8;WL;Du5DqPe)%r5-b)T~XEhFL~@uh34TV?lJkeL?b!|Zr4wcTxlp2
z#!Nv)t1T_{4}h`y2HG_+g_$o{Pn{l$`OJ^au-|x%Ru}by&I;jDOyn4|*A&BSZ%FQb
zBw1dSVu9BWkvqs2+53w=R3@IonvNMX^T~IfTX2Qm^mk+PCSAgjy(4IhXC*Ix8zbqJ
zCEPnIX)OEDE^2!{hOKhW#Y5W-(D+qA7TvGXw9bAMH$RijFuo6s#%tU$;w&b+h4B>G
zeR!PP#8rCyW-sf~1!KPsNiQY5{%1M;+%^-=FAu@&*awMoBVqjUqpWYX;5rSSi>{8T
zcxo6+=4EHFASIOY2P{E)P7N!aIRWiaVUka?r%;*cdRjYp8_QDkmj%_0V&gK5a51-l
zlVK@(ktZxIm(al9pU~q#0-MtG0R{=Vtf^uuEgidy91@PxerW(rQs@J>$qU%Hw-W3+
z86#8m>P;rUL-D@W6L;&L^2;jS;cBWWTtEh}6Z>B6UtJ4U9?M}qZJ<U~UpU+=rC-L+
zFv;8t#+!2__U#MNvn7zvJUfNYtnuItq0#8G@*>N_PTn!@D%|}3Ll>1#(EOJzTREYQ
zb#@=jzHTz1y3RN7yr6~U&n_iJ{YE67&Xk4d>O$ReAeY`T!K~UHRBki}Hyq6P_%Ykb
zyYU6yIYp34zZKY1^_=;<_oR?}BHPqa%(pivK`-YsYf#R@9Kqn6`_7%ZjQmVx-;d&f
z#DqRdAER-fe8JaWk+jWG2}MC?vHF||^=xh7A>je6_Vpsl5d69!s@AwZCJ)zcePum2
zuHlZXgj)ThFfDjE_0KFu&We%b6ICWrb8)8lCu#If<_pIi{w%_G4zChtRh_hfg3EeR
zI2840mti8lEjWRk@(A{I%yVo$P$RLwm_{zauVI}X%WW)NNcX~N*?_-VT=qPGwuIcI
zJA2Zgb$b*0b?Q9bPp_5D;UQFC+aj@1+#_rMGno3G^+VpKD5`01=F{g1r`;XF(ESu6
z@p+)a+wO*<;e-^&8&@DgF$i&et+D+12U@<eBY)S?pG>N{h64On!Y}h5x3ynFhFWK2
z#}r!dM7{^+413NJLmIJmRVIF{(?e`8v3s7fR(7hZF{M4}Ay_k?Y2cut*e^|DsRh?z
zGb0J@T`mci>2F-BQ-*V-IsJKN!XM3<K!G~GT<>W%oL{@pAhUStz4t18Tz7}A<hPUD
zg5hxd-p(_WlGxo(FQ{gRGK`JNF|qAn{sT?%-Lmo}Cvn#6==)0S*o+anLxW7W{N;4x
z9zB)HNzD&km#DjJz{8U{`Y#%Uqx_-^Wi8!GZo0#He{sGXnEsNg*i33{m6w)&orNoH
z!sGhuDoyIU5=ni_nW=mXS=#(Y-GwGhpWh^FtT=(Jx0$S)YY;jF%kb32>)0B*0cV~*
zfMZM>3l*8@<Rf-)nfsGZjY_4^$-kKLw`x+&IzoCUS3*B#64g}4;c@j@sJeG4IIa6z
zWVUC+Em~xpUt6Qt{359>xQd8A8f5#giYlV!Q-0nP=1?@3X7ATwk7ggn0^W~?_vr$Q
zpzF*=Z#aGbUMsVG5lmloLrFR8y}rW1-Pn+5jeSQHP-b4s4+sy=;UIVH?30XF<HnJ-
zLmF+YiN|G|XndS|LO8-k^3SfGG=5Pr8@?@oTthG5mB=2)y7z^?q2Opb50E)}P2&C3
zK8bz2Imv$s$GP$$u-_y)wqII#?#z+MTi3#FI)5PJW$vixdzl?JE?~abzp?Sy0Au-L
z{@Fi`jLjNooBUzCZ`({eQ!eo#Yg}<NXC54bjp?hzkJcI<qqtM5@F-FsyER@IUvd@=
z-Tuf{Z&F9P{z7{DF@tAs$dWiF6=GIiF#ECVBz+!GBAZ+zCNFyPU_ACL;^+TI{Q4Nd
z3hRQAC)Cgf<HgyoADU;ZL-%jOk@bEXY4<T9m)U)BBzPci`6gz?R;5@p>Nadu)acu<
zGCJxr6u&8+wC_L4tEq~D>}V}-%vnmhzXU_}$X|A6R6EPM@(5KwDltQ!adn<VyDy48
z&lq=HZ|tA|wOrEKzEx!LVkv7<g2;G`Lic_VG-~l}mKCS0Z{Oxm2K@?PR`3$CMiYt~
z-%B>m{urz7NJqA%Jzp{<m<DTaquUBkXrH-1JF@Q|*D&18Rpe?|>Lvqrb9f8pl=YL;
zq)TbZ%?aqMTfw{E{!e`O&Y_|EM`-A!)9eP(5vr>|&xDDj;4=cZ!>uIh4puPU$)UHl
znjPq}nVkD3VBPI_WPMpIaV-0z|6u1+zIpX7sLL1eLoMb!{B}8AA9;n$)m_jhNL^a;
zQbjs;jt94x<;@${J`jA6AqW@zSj}8Tq@9=0*wY`mTI?FSoAw=3b3C#A$Rs>T_oMqi
zIx*$1L$LQp3wwDaglbI0eP?+VZ*e`!mc@&mMgOl9yMMRH-^z=uTsw^m&7farMl+?T
zZ!{tAJc~4}g5kXmrhGnIqTU`W>k|41JDMW-GL=iDqj!$A3U=n~XG^JO;b^3bZ<5ZX
zPdG7Mu-W&|=Tol`6~&E*=4~Io&2fl$9#mvkKdq#kaKU$oYUArXezMx`>X<ZUH2#=6
zuu(mZklzCj9P8%{he%Z%(FumP;1+6cx<wi9?ReRx#jJE_63cYUW-I61LUyA$271R)
zNXicy_TV2m2gE}z@eqx>b5c?=?zm*EzZ5a@J85iKf62GbLCnRf65(ZulE?OG)V*3J
zD-L(1gQl61@_SvSM^`;Y&*4{LEasP&c0}|1UMHDH@kX|6Bd~bKY+0YD{;1aW#PH^0
zu&7L=fxTK-@~W;B?;|n`q3h_}%rLSU`ho>|AH--|V`kC39qlEbC0|^VnBL&S_%iSs
zio%}bKmsr-J6qy)K!IlmJ*A!+zflmijP@*Qq#Hp2^nFb>@_H_&3l$Br_Jg-^tYM(c
ztLQE4?r;=1jwa>XS7ElXGv#bO!Y<9uMB=!5y8Akhebt)C0uDdL^=oGMZ6Jl^>~?<o
z+)7-zu!nsMeh;>25*8twjAPRI!8L*0`sH<QIOQzloz5`joO+hG&4`96+F|BIZyIqn
z9y-p==y~Nel_(X_Cuya=PTdz&tDDjpU03?MT1~2a^g2zfS17Pq8!njZ@AF#r7|2fU
z*^aX<<48OGA0O9uE&RVKOMSSE4g{QmA9tssD0f(o=`5`{H;Er>cEi-^Ze(scR(5CA
z5}3XA!-Lhj$Q3gqkJ3Cyo7Zv2*)JLYeO~k^?jrh4E?qwK6RpFS!d20qbw8>{T2{d{
z$EZ@j-=rw=Zw|w??urGA!c*Z{vWFI|K1^4JoIrr!_1|QjG0L=p`Gj<r=6)Ya)mo9X
z+-a8Nysj3a2aRMNS|?~v^GF<XNMb40TX=K!J~I4PiiwFqc$4%Qi3Re~%+LU+@5`Vu
zmn{&bGz&*9XVaq@a<o+^ib{tbr?1{_SbT6Z|5s6o*D9}|^zs#JX^=?5H+sVSt3L+S
z$Wv9lE7=Yo%m3cbLC34ZOtEw;jQ0zVm*qhm=w8ikmk-2|Mon53EY3DvuffJ82H8PF
zS;9<hwlSohsl4KRQ1VgSc69}*o<Mz{EIz=-mpqjB@$?ryv{9RoA>V<Js5dC=|0=h;
z>UwG&>cVF0|6pm;w5XBdNN0**m;QUgqCza$&z*xP+UF<@-k*j|OO<gaWh2Y6?O@w{
z`jUpzU*0hwGJkYuf0>G}3WfA9go;%I&8|30Mx(v4_ywV+&7E?`ScyIFJF<5l$6nAI
zJpZzl8p;{e*7b#DUU*)%LTAz%k<6VV$5N=@Kze%B2QmwPcBM{qdNtCZIqe}e3iqjj
zr3VHIM{U>1rBHMAVg|R5LSfKE_+*HT-S`AbGb`pFT365q1y4G5<{aO6{}l`Z)tKhe
zo6z?b{_&R{G(5hVJ=1XICquhK&-o7NRVPV42Wj&`D?{<;(ks4hNDz)~e@uf%mSMx%
z3Rs?8gn94h&<M+Ocz3Zcg)YA$IEE|m^-v7ke)<7w*9bTF^0|1DCvs^j(WIQ+Bndl}
zkJN18oKG;PmWQ3O@Qe<;=EtB|X#fjsb(W+Yn#*m^g|jT9PMAA*Ca<5ZDDJy0yd~~C
z#?~z*y^+h<$^3S_N`A|VpQ=!T+6~$Aje+nxGl0L;xzGMnSW4=Xj_BY1E_SynySc6J
zV9DcDc}xvWWk)9VVkv2+Z2q)xjOh20P3@_N#Yq}`vi(L{eA14uZKz_U?w)A$wv-)h
zV6vLiiqfwRD<F3vNoHG8h<3RTsBoE!<oaGzsGpIS<tl^M{T?t^8HX9qdt-xBB5$20
zvMR~vF_-m+?H^5y+AV{v>t)tv*iLF8i*Ud>2_J9NvN@L~z>+|Vvh}bqK^G%dN3vg|
z9tk_0V3H@ONK@p?*o1j!Nlm*GeANEX)MO35Wy(ZkoL;8Cr@K4pzPp9_2R*s*DjAL{
zY(N)}GbDN2mvx`>4@QeWai6mp@GKgNz1qixli~{6i(ez}x{U74jljA`>v5z@0Jb+;
z$ZRe?L&L;2{Ywjk(|DpGi<=w`eKiw4T{(beS&gEg-xb(5sE7FWv`dN}U!kCeJgnHD
zM24?=q57`qb;Jx6Y=3QpT`)pe7jH_IE5xr7bBd_#hGN}w(BBzM4)5e3H!AenUEfm5
z8ZU+afFJBw<4B${)feyX+0(bJdh}#l5Uo6Mj}DA!<WB8^XYenb?<uaJ%D?Y$zp)90
zqI34=>kbTL9W1B8g<5k?(a27*q`1)#O???o1O~&UaNg|HlGj-Iu7!4QxJ556A4f#W
zByl&Jgl#qVnPZj6Y7X?L!at%jSlo}!+<nM~s4szWrwgJ>8bn)GOL(tawJhk9gn!bx
z#vE5Dk)GIZhMFJcipQUU*-XdQKJWR=A9wQ4{_$h#BVEwqwHk9w!g!&X726qbhcXAb
zu$`_tl=nIwkh>^m)Lk%QWidQ=iQZVNFQi&dOmoy<gsu~fhC}{HZnUTU+v8y}Ifl*)
zuiMP;F6j24UN))xFLO7%iK%Um$T4@1<YvAnRN|kLx%&fL_E#)O{_>KBIQh`Z5m)G%
zfs8x-xhr}4JQPdPq6Ft{GrCN>fuSB_fB~sgp_#@FJYTR!cF$O~w3x5)4TOWwQ))Gh
zV2A%r!JUhQX`OU1D?yV&zV*Q9;%stx@gIgdeidEE7TM?QZrC)!7N1(?qxZ==tlm}3
z5{EZ2xhtL$mp*HtRy>C0&D+hj>lf$cI)u`afA4vqL4j<_Q(vrGP@iuz<qSsM3nG(N
zai(*-$P8b7r@!92kb3rwn6YKE_eDJrJY^!i|FRv8<8I@T>tOt{DCaE~%xUe&NbD6k
z0+QxXZP`rd1xsn*%pug28v*ZaX%x77FS}&+nU@5e<E<-0Fy?0w9IBd;vfG|IdseUo
zuDN`bZ8VJKE@5#|qy9DJS2XGKU2^Kvo$MZ7!|Imfuq>NSl^-8rwPGHZpL+r8TCLc;
zQF~e6wS6$DB8g3_O{V%mg(Mkr5rMwP@aNG@-bv&u7lrPp(%4^;cC!Wi$o{jis{D@+
z@tBGyX8!mX+De!1Y=WG@NaklEI-YyWB<ag0q0^Ne%yE&1%vf_Q>-pp=@-Lo5!Yv&-
zbZawhIU$(sxu@Z1G=ui&+R;S4Ak<F^L6V{;8zg4L-@K|Z)IX5^E=p(l&YprNdzat8
zUnWahSxjwZ6>Q}`7fIg7c{nyD3j0+G;4RqeZ9l~Qa`$hRIMsxwPCp6nDT!D#u?q!C
z`$FlB6x+`xA-JZK)H?Ju?=p3!=y^Mnx5+FlHyg$pC8N+>H=lw#%|Ud=DJrNDeiX-O
zxZoK1>sIm6hdeOiKopDhkK;EU4<?!D*;n2;#tnDhrH*O7@V9bCw6_9stOZLs)B`7M
zL>4~Kp4k~4q{qsu*;}oZVs5L3LQmlElPJWUNZ{jjg=f89JfA~8@}vjnp`m0%jxQCY
z=VyCjughu7OIBjdlildnR#)Mwea*_B$RjxH5Dk%RqehDq-u_<3`@S2`oz9%1i6`tO
zt<o>>4aml@E)ncT$6=<^5smY&TqtnZJ~Rk!V6H(N1(ffAamh%gRi1#q7E+3KR3W{J
zPqOWY*HcrcNoaoe7D2z0aD9*t?bKT(GAQC)-7}rs95+*yM<|>AV-bDdq9>UkKNS~L
z+%eCx9D_bypuZ8~`CTY_Z!5B3tJ6ukt-_OxI~Ot8>yzaE?;57=se;p3@V3dC^kTB;
z`nCRM`&ahCoqz)f)OwNc<Pt&J_s(MAcnvI|p)zgN1CpG6E__U_fh^2sKFNhObITZw
z{OM+M_>tX)@K=k+U;TV&Z2pE1Me0~LOPM;<mC-Zw4Ew{xS^oQW`j#dyeg4K9ZHiv#
zTH-|}n%$^il@|pTROAnN|BWB`5=xWH!|2zY0-Alk7$t%2@cwle5sfA6RnL3;Z1Ed1
z33O(AZ^TH}ym2PGv6JapZY{EwbV82BB6=?uNZRk0<GM*_C_fUOvu}m^15RH;lU@?a
zMBcJFt)7iPJ)Op?^eC7oZ^WGc?#I-u0@PRDlc;w0q1S@N@HvRlxr3br*F73i;o{!%
zS)LqsweqEQDQv{EpS<7S3<@z$g`C<uzU1;qbd>+X{G~ssR_-Y@iZ`Oy)C|~OD}bxX
zLVm3}SA4@R!t?Mn7EqUm0KX87ep3W(Ylr*=7EOG5S1&qOlptoMrsQCLkuNIxs_)-t
zKJ8Q3B*3(X*?Vg<W}*;4`qhtR+dqg-<;D^=YqdXC-qWI;!%ySrR)0F)-2;sqLor+T
z2YoNmB<nE&^lU{KovT}lV~><sf&NRh4lS3=i}z!JhMAE5vBhkk38dH>i?-9<$tl2(
z5=B;F+1@OPi(V%*EILOMLn86)SS~dzJ5Q-IGkI#tJ#<eirGDT2adSc@Rc?%+#Q&~Q
z?7icNetaB9a<rLy&x>rp4L2y?e8)P?FJynlR8n@FnaIOE;g_TD@ky??*#>?An#-2a
z>a++f_1;WN<n>sQ$0f=uTTF)^t>Yi-6{N2Vm0>h^36k$WX5$W)v7eiCkal}8U%Jhc
zm2EyI+`Quu_PIfJKCyFw>IHj=TgU{srRK}_{}Qvb@>sU@vm2@&_o5V!S*S7ZgWJA0
za4_Kr9)+f};Ult9a@hrTj`2{{3gbKGo<^&Y3G?4E6z#5qBx_X`;l~y;WWIhQE1sK9
zebv3Wn^8TyyB<c;nOe~|uIEo{)}U~P1Q3)9m)o~lzd~`3bD1wwFIGXCLNvGe8G__~
zUlA~J7?m`=ASe488mOd5iE?3hkQ;-EsZA2^XDhhD2yOa(aVCB1-krui{Ki9M{s_Fb
z69T#gv&DpoCvM9WqBde!bOfHA9gnnsU1_#KE$L4_g@K*>(&lkbB)fO2G2=~j?8(|(
zUUYIb-!b_YFO#HERGl@<>RNc3<zXzmt%(uNM^Sq03cp)%kIEhh_Db)qq?K&Kyd(+C
zclI1oTzg&kxg&7xaRkN;K7xTd?V#E`O!jl(t~y=$+1~$Hw%>2IlJZd)e;xMTWz-U~
ziu&vukN#>=)F3a#t9WC{x)fjZz99Gx#|NV3&pY-@KAG}!9tn2+J(SNY;g)sjd4(Ih
z7If>h0j_~{SR8N;Rys}0t|%67y$^xfa%f%3TRzE21KX5(;pz0VI518bDUWI>%r}K@
zp8#2IJc*nPhNnv}uzMqJvXt?{E!lTErL4M0%Xhtki-I*3L_CD@g{^qAxQ&KR2!g`b
zC&J?vN$Xxd;SUO8s5Qk>WCy(Y9=}_VYzu|nhw1oQ=Oy~2=XrU>Lb|ot2!is@-sbI>
z-Sg?f%oOq|$8$QFA1PpZvl7|w7f<Q&w|Hh}^@3MEzrYM$+cBLb$LY|n>oPx!NYqXC
zrM-dyxwiT<eD+Sq--TUintLMlOj{ut9jaIm7SjZ$ztL2arpKaPUG=vNlA`P0omBcM
zJ72!D5$!NHf_}0RId_u4dz2P=%x-6C1LF9~=P8nZ5#l^^I1};pC8Q2dT)nHnOvLVg
z&ze=N^O{`V7;7Y1HM5)s_dLcjH7}7v?>dsxp2Hn<btNXfT*<YYJ3s!s37_T1GQHrz
z?0H-UWEowgi0dMi?|Yn%MbE<GH)CY}`|FS$aE%>`RL9W?7g4;X3zILn$#lO?M9$VN
zlC@cFEa#~%MYmpt@~x4y!080|iu-JEbvZK5#fiN@58+@{koKQZ1?Nvi7_qdL^}U`(
zsd2A}f8Ieme-9x4HM;DexYK04lc)JTuCpZ~Q+#1nJg%<L<O@<cRZGSquR?SX&CW5m
z6{YyH!H&MH&Jb=QRTM2b#U@YgitX;PqWiN3`47%e>L?A8TQr(?o?ML3zddMokrBn7
z5uDeH`55G1E%{=<mg&`eLBr^7q<tq9JALCQ?VAy~54OaTo%h&OWk+^j`3dB@6|gU@
zyGc&Xm~u;=VZil4G}JJNir-wp&&X;tepo<Ff9q*-+)3&@LkrTd7+P1{jeHKIlmDv>
zvag<xa-W?v;4)|9g==Guh8L?Sa7M$%q0l$gf{~qIB2_7|1M+Hco+d9HJ6;FJ?g*Ef
zLN?3&e2uEM#beZtOoVq`i&G|Iro3b~f1&b8wsD9Dd*`S?KaaFYrha`X7)}yy>9U4h
zype#y4{o^cJQ^jjuc-9<NA&mK%@+53j7V)8{$x%VV)rkgZht<qBNwgceg9AV+?~;a
z4f2fq5B}s&G!OGv-eYOrz)lz!w~RvCcCg7(RZ_G6O&wcwP?ors)~ELnt{2f0S5c>i
zGLZ{B5yEC)7w6q0cl6cCg0c#~VAT6^Hf+rfGPT#Ds#+UtnDmf*MAtsJN)e%I{i(lf
z32m>MFIzmcky}hTO&I}ikfgT(swMV(-=(+Qt!^dnRk2&*mvxV9cgV{g7JWb;?-|VO
z$2!{ArxaZZi`cOP5p4OQBiJ6^j-1oJsI%2Wz}%%sJ2jj-4-)6~tOb0`o1K(p^Bw>G
zt;gs+G5AncN;an(*q8nDa8~mHwww#c({cN8-0vA(e=Omb{>&%UL-Voq^=TT*{*g^t
zyzGJA1D;p2lWm%O9mDR=M@g?{3O%JvU!|i_`=A`%^RCE@*P0?O^@QN<{$mG>f?+)8
zAj&V=@h^%ZC$RP$9m)HGo-cP%yR?;~`yN?FMt|9F)vav2|3FA@_yf-J$TSvr^6a;8
za`M5J$|$jy*$R1)CG#4(g*vr`lSyO<1!YWS<9pU1^7&h+Rt-UehBHahRH!9B6<<~Y
z%y~1rJi82$&84(_OGv&)jRcnVp>W`naBTB<wn9IYWpz)&%CzH%QMpXt`lylgj5}qR
zF@1+an@K@Ejs2Woi>03~GPULi{_Jj7c+8!FoaaCDE91_{GI|KdP5MxggAAePTUuGh
z!%c#>wVGB|uV=BRgK=fJKgntJ)!#8?EzGv;Lus|}?y6oUw?j27p;1QGy9&wpLZxKO
z^kZE9hoA6U4}?*&=mQyU#pqd?ke@Y>l<sOHWBxwG6;_g5s5ew@?S{jpX>jbOCX2sv
z2(|qcr8=V>pmM*nZ0fUd)P4M5vY7OaEe*IOW>z;zKEJng<PzbcOiso69i2(1rW%>I
zf3U94JbC!$By{<5my`Dsvbj`(nz<daevLW&_50y8W?{I@zEitw$E0a||As8)=z53x
zh^%41&zsR{csb3ScM&!6RdoB87oJ62<BR1B@i}fXMISPyC%02c_qh-K9kUKAL~d~H
zg`rTp?!?+w<<Y3UizMkymnr`BaipAk!F>mQM9%kg96Z+wExAqwXH&kASwbcjif^+;
z0a4%yf7H2U&|%%n@IIHwe9G?g94{L&ciP2gUL7d-@NXz-!)eMs^_P`K6)@!}MKbu$
zmKhjp(A4~Kbm7c>!RL}Nk02AMD4&LGkSjI1KjWjsS;g$vai*#5FP`VG^m<qw-{}y6
z;eyBOZ|Y8cdJ8u6#}X_Q=Q8v7d}u26VFw%a*bl*d*=6*SoyrNITFYp5Hn3U7Ra;Pb
z?+h|7#c;T}!_?Lao$GT+?Sa3nev1jSXir6|?q}LB_<MDqcH@a>B04>=CX0<x==F5~
z<i@GszHk*NzHCM1`Y0rR+laqj>IIP{Ke=(xYuU5U+v(K(4&E><A%CJ@8(mqi&4P|a
zQt1OdL|E6;!|}P~tacjfPg_v-rQv*A!**0f-p3-dN<es;->8(sw&^n-XH>HR{YI0q
zTQKrgt5W(FFI2{s<5X5Cd|l-5GIImEEj!0Q){Z7K@qU`^eH?w(Z=g@jI@I;|Rr+?s
z56vF6sHy)1lc+e%esq!YHvMMJ@ggHO*dCTB;(z<NGTRAhXt&j(dAov{PD4M~3@{V#
zmw7nT(}`C-Ol6}F#!EE(BKV}=d+<Fs51m3Y&~l+mzT#1H8nQVE)fQ`M+twJ?Fljej
z`ug$o7F#C0IF*d&&81W2s}b?7AKYiGq|~+w8fe;ztkPgwJ<6PF*N+qX#Ufm~bq8aY
zTfqAG3tWiUN=7~>=$2&wo*yu#w%cY1Tv^WSZF*wG2UEIh=SfDb?=aLxblIBq$=-Sb
z-pX#_q<uA+?Fg4WIdhVB#?`|0-F9xJ`iR@Tk|0edPq0h>F;>$Iqm6QS+Vv6}wr!hm
zQWoQ@`FMJKJf6a?JcsMwDdan7Gu<j(kJu4{k*Rr*&)y|^Y_mN0A;S<-&^S-aMLx>#
z=S56Eo+s{U1_(^JLysOLiW%e>-t2i4*}renZp)p#^VF}*?%qYL)t!YC6Rhaekcp5c
zy0ho@xlqZSKo{dKLZX)`x@&qAD0hu*O|!taGQk4=F_JmXHD+BCy3wjp?R-(#SN`yK
zFC;A&b7SA5Xb9fKlHZ-C!VRynXs$X%?zltYK11kU^$S|=swfRC`G`ZqPVzp`mYfs$
z#g#+<L(<w+RMle!bBuGMK@;Q(`n(uR6D=FCwre;V24$1==Y1GDNC!=kb%?Q4lBylc
z;dujMA!lqsl6PK|9bSRtPE%x+oBWZq#fd%+KStjY+Rz%ABHM6q0gAgRNso`c%2I6K
zLBnSez3tfxwy{$s`3mRR@yv&?zH~pYupo**3^~Uu0=JRQ<45@V`w8DZZVk-Gwa|Z?
zI!k}r9Hj9}yJPDqD-^AKBbeGN`GC^{c$-e3O!aFNEgq&+(6r<kmu!y5y;YU;H`a$V
zEsg1r>33SHrA2|E_ZYj`AK{}HLjFn{*Z(_^Si7T)sqMv1lRIpz)@587Z9-#jwn=6#
zw`Oe71Uju;PH8_ADUfZz^09*{(!UN_%Ib)<tmn7;8_@2FQxV)Rh1S@HP^W)w$ol1i
zl%6Sc)Xi14e~RG6=lSCK(=>!%JtHZ7FdDvtpQGOsWtMz%J{4~Zf%mR&vS#az6u<Km
zR7Ku-@PU(TQ?WN}Lt>d%<zC@XyGWHK5rT>8N0m++`Q1H+tk_&aXWl4Dza_jvLaPCc
zcDm7n;DOA#s;^{L{6`jQF%D0CgE9NbM(Q0^1o^(#5azv)+D7Lv=We^X&-w~v&6`iP
zDf{SDvo#{J+h8(m4i?zIWuBk6lEe8^SS~UqSBzxXZd1=zXr5+s%#{kBW@q8l;j@yT
z`j7D1;uih#(Gq>53H<EE5L)dU3YQ^Qc)O+s7ikW0h6&~ht3G2z^)>8$FTNEj;biLN
z0+kEG;~sv14m23UsBSerxUNIAsVig`ry%M?GyC}JiRk{<(5d=aY{&W6-22>Cs_R&U
z*5FC(&4meg=r6qP=aX^y_)WMgzh>XtO=<P85K7z|iNOxO=#syRB3FH2QE##^H+vCU
z3cGSew+lR8?FK?36lk<z0;1m9^KKrCWW8oxrKfBAFjc-5y{hw}KKC3qb4X=JZObvf
zLO2X+&e7qR6#iglJF~2PLr({6BbVe`Y^$CPUi6!YcQb~{-1^sZpZoWjEVqMyyt<kW
zj1r8EZ_fp<Y!_7G+9Z>@*I<OfJ32JN7=08^BWmv^df)FUjXFM)IvKm+zz@#fNIJ1W
zX?L0Kx|7^k>jwHdcjooU!MO0pky-AmrTq&8yLl1;QN}QqmCE|ucakKfe`jqcPV(!H
z)1Z@HjJ>B`vx&=o%#x2z5;Gewobr5yLza7~*5g0?)GX5vep}Bz_J7Xij62Id&Dx1M
zV*}B7Xdau>V}tMu&Bup^zJh@lMme2cvB4W&V&$F7k_}T&(y){$82t1>*&ZXB{&E@m
zy!i@+;d*4*IUH`<YO=NsEAeK>AG)BF0L{cRv@_uxw~5(6OTXK&WJwtxpCZb!tzRIw
zw-ayhX(pF#`w%sM7OzQrNN!SnK4Mb~Yq=GR2L_8VajKH^u>1i!QF)(k-%z3N&-`g(
zl<;DVhv=@~r@7O$$$x((D?1m#Pd8S8^+}-2xtV;XU4OQEeKmXZ+7J&iK0sTkf-UUw
zn13Jaf=aE;XrH|ntHizfm+28wwY&*ww=2S#{tty$>##(<j_#SP!k(ksaPRGKblD&k
z&t;9M`TQWw)!rf_m>~8Xi>SJ87g+^WkhR1LRUa5p`$+n<+8$;)b@(wdonJDUNWIf@
z`5Q4pKjtu&hIl*RT1h;PoxX^?KJ#c_dL(SFiF}H<BX@BcMK(vYWG<4|e9D1OY{%d#
zL=R8Gvegl6REsOGkI6z<S|FUPYK7ioG40d54buR5()%`+5~c5<UY39*)kn~!!$hzy
z<)!m7Mb^LKjBwxfC#|S-Xt^t5=YP+bO?W($?D3)T<{Fa1-@-9Bxd3(rI<oyiC-{Q4
z?<{_YF4PuZVF3yK`SmI{+}m8i)|$tXN=-CXg%foiSjHwUx{s=7f|0V~7Wz(^g&}ec
z@De-h^-tzwOjx*Z=X8fgn_%xNkHIeMhY0w#iPCqTfklt&xa)ip&7X_NWwDqky<LP+
zA2i`@@`KG=P=iBrOz7B(elRGDmbrL*;De8bp+)x*dzif!FH@({f=-iZ)96?zDgGs8
zOTm=uF%MO{N9ZrGp22Vbn8z%0>u1OA2uIYvQ+RtM08PC&<DHJk9T!~SW*7WO&u|3}
zd|XO}w~k3n#rwP=SOKq9z0l3{42AfolI`Zf;=I|9xm0JcqOwTbIdT?{lV4MpiaOSE
z;T00^zQm!ZV)QBytftUdB#IuYVyimtPmssVM}P9=H!5JAa}51xyUp%pnZToBHf-0p
zvb?UH3La{X73^XIn(E|9Q9Zw+V~`uEFMCW~bgk*%@NG!!ESSS7*RZ>=ke0uX$MpB+
z$P6u|M4MS;pZu5gK0Omb#e1<Q;V_dwYA4Q2*YMv9Pi8sFk!Q*+WM|VhpoP^+tR`Ba
zT-}0)uYSxzcIDIbDs!5gbsvwYgEh7}Av{V!!rm>VfTC2%+1hnb|L+ege5XjH6d?8s
z-|0$5H$>VL(ucjNr1MIPrhhZ!JH%Wtar6wldNdc0f1aexch=FX?yk&x%YMOkh>}1@
zl~mF{@aOBDD1U1(JRja9%U$;LQshsj`6!d;95?ndPeGKR`XKYpBd(!e$s@g|W2|~C
zJY!#Dw9QcVZ$Tl9&5m;ur3YBs9oW$0E#(dNqR4jvq7Ts>0n;C0;maYkRilccHSW+S
ztL?J0-cgeF9rdt$;ld{#@`U=Q3p8ryBP@QomfhIm#VYN}B{^3|$sW~}@$37~iElzY
znqKFK+@bI=_ZK}9T{l>+eF(>Wi&$=xJ4=tzr$W;pq|X^I`uQ&Wb<8kycW7kVW4#dI
zsR5gX7s#UMF#Oz40wN>w@ar|Er+i$pU17R#yzFIruGY)EqMvi?t%iJZlLay@1>?wP
zo21V4Gf570a2m0PMS4#q|JZ02rWl8*heAlN=W|*;t_&5iS#&Xs*!f)%7%#g;I^Bcm
z?0G|WbYvuX{P#}(*WXaqzFUiwv{s8QspxPfMxo8u2f0%QVsgq3EH=!f|9YuNmsUFB
zn4S*3*jpnhvj0ZL&MtiX=v~ZV!DMVvh+-d?uEFBztMi@nYboexI%T(BftKYj*0@Ou
zMRz+sQP~53znnqaR|`JNOfVe^7r-xN0fxz+A+Lq+F;_!EX6}oTzAumOa`R#%wr&H<
zv7*H3>6ov&iz+vt#_Edq6p)jTGlqweks~j?ADmCB<q?#$SG~Z-AVjjM&4cDVb-|}M
z>9j-aw3p%^rfz*N>3!9kP8@&AR@x<U!{$|d#={nR(=ma(XUs%?!g4yA<$%UJhU|iM
z91VNbjsI-U#F~jN6lR-@Adk^_@_P_`+)q&AbsG$HNSEnu7u@alTd?F(EUsQNC-0i$
zcxIc5!8&&_ez`T>d-4WWJ7(ay(<^c`?MtZ^S77N`Mx}F>vF*VV$gHNBH6PTWKr3H1
z(c!l2S8_6Zx~r4_n0FLcnZb4(-$!!eGH|kL4IW>~g8YdYINSV$w9+%#wjn!_G*yYz
zhVLV{U%|Y3R0X@1zK@ztx}jfB75cCvoL$>i%J;lJL8@O~AV4{gPk9)@XX^YX%Xm3b
z_T-!~U*(%ke}8CW(e+-i2p-A4jFM+Vbe2PT%4RBlmx9Hjulde=D`;*a?VYcS^`Wtp
zU=+iweD6vOhjhSpg5V|JkdqEBK7;O~lbM&x2KL<4o!bVeke^O1?bP*!%6h?YS*x4(
zcW^vizR1`l-#a`;<2VIGoI_A+2siU{r9)K**!aKaA)8N7AN`A5^43uO>w7#;Qy$Mh
zyv2OAGTz1N8;kq5hi{vD3^nPNe1~~2a+;SwZT1F&dp3c>tiot~Sr~R&t>GE-Tm;`u
z4Ibrx=w;4TY%=%b9gbHp$$16F+sr2Y1;Xnk>nyEbX~UeY&(rjUw|GH_3VB2&(Ad7h
z6~F!i@`5W_pI%>hN7H0z9y@_8&Fe)rPt0X%X3z}f`Rv?{u}E!OfmN#l*<gKJO8C|Z
zQ+3WF<+M_PlChk0jAvg;)K`|M<oL1AYC~k-sx>M4^AOopuR!P}mSVDNHCK@Ar$;g;
z+MqFn{Wa)|eWxxUWMeB|^>!uZ{GLvud$ppx%RMqY_l`|l=#B{Yb<AtpGkj`FgZb@N
zCNXG0(|O^r>b#4-?abu;e%O$o$s+3CIhg8y`m=Vg6q(d18!P9<@v1sS>71RtuqA#V
z)-Lm)WZ%=|84-_`=5su|Xaja?2BI{29UT#Kg!i@W?A9iG=CCV?WgS@ppJ$H|RFliP
zj$B1&QYtY#;V}k$&xdKE59!zueg0C2WiMBtBW)8oZ+GD({k$n{@lAX+o-f{iIqZj=
zmrTvGn`Ehr_#Jlf#&DlI!aH_}>W|66UMWMStCt4VlmQ~g^AhFe=6G^H1_$QY!n-XM
z?SB=e?x9y9=?G_3q%C^hXK>Nf23`Eq;iK`441{YVtTqCLLCs>vyB}$L&rn#NBR&=$
zMC{6?2rBr8k7LBV@A4wV{@jDN3qM1(JxJtX+}UlhTbl0rl68w;OPwOCi7wl~L1_rB
zoOVSPtbUEpiLGEubl=jO|IWzbKK&%q_5TTmvKEZL`O+!B;c$!GF0*03X`i++ikeRf
zXKQ!qjcMZD5qSkV_eH<u!zsFHJe4W`u0uh;)w1y?q!eDN#vCpsqrR&Isx$nV!L&<w
zsPu`#Zk?h&f_1D{ycL^I`3k>AEkC`#j%}WyCjIp@l}ZvS_{gg5_`3NlJjOQZcNors
z$MS91zFishOT(e$JDnE}%*k8Ia=1&At-h;nGc1lTfTW+h<bI_Rb}Z83>XRRl>vu<T
z*3zK6#*6uzLnGl+olIp*qEWw1Nt_o(Q1rk8Zv7{oFZp(tD-27cBXgptT(Lv=oucVy
zluUR?*5Td=C!~$*BYksXAr5p=L13@pSUc4NEyH4DHAj;9f6ifaKJ+ZYGqn*{)Bx2C
z4ZNB^f)ZClk~Q`g9{D|hU%r;Cn?oLb=$*@^o6itk;%Yvz-(ub|OOdWT?=BtmLz~UW
z6Fwxpb1<R~bekE+at{As9_j(i?_nJ?7~_kzOCG}1)0{#^nelrw_u-KEZ7UW|qu#H5
z*aYL}q!S&&v_C!ITc)dG%fLXKU-y*%oZ+v3$TX1^rZe1_HV!`XFSC!?zGN0VfgDCJ
zLb9<6b=#0d=9$0A`Qi;)uxB9pXY?qTRd$=gj{U-3uLL?=cvbXr?~-xkba;FZ<%X|?
zi>zZ2yD=nz4&1R|4Vk(5gTsFDhc_=!LrEbGJwFXE`g%}rWpA86Ie|v70CXsRhxyeV
zcx}{|^?TS(KMxH=$JZv_wbxTbrq5^I=frP*ND-7i#!^va6o03b%l>|L#)7B<3d)Wk
z>ppk*+n;yQq`#ltda*=&56<#dJ3pGJI+Xq_3`CDZKX||ZSBZD#0~(_nCYauf_=p`3
zsN5qMu_=W}GRY;=p_6%NZ6v8u0v4Hn#F5f}JnlyVopcUE<Jp5O+S8i--H?K{B}<t7
z_xXr&dWk~6AQ<h^qOqI((D`ozyXw&!4!hU$O`Ez(J5WipjKvwlDH?-cjzdOqC7rjs
zgxLAs=zS}T`rJK-+Suh3{@0fK-*cs&pZzH}v0Sol^EvV}$U`4@8JZfGP^a>_@G3K8
z%ltQCPP>LowM7vRR$XQq)k|O)ZH!$d+7xMjoT;w-4CA0nY(hXVT9pUTx`I?OzZY|?
zeampmG?p$T|K#7w9AUfQ9=Ch*hHZN>ihpc9M%7o%XsHsHP5*kGO2-_aoV&Wz*1wn>
zO3z6itQ$`n%42BCKYzTn@I&>eQLJl+IZ9@3rc+L4Y(hggJ%2R^;-(~X9DNbLQ}Su3
zt~zy^zm)V>@26pb&!8Y@k6Dv7uqnAONuKnRX=|U(pH`S6_#NH2!^d^F-0+tAy7v_P
zdu!gSb1DL+U7(DXAU@ef7&C3|z`i0-*0$_4J$K8*$?08i+`xnOr)?6)pmxd{QqL#v
zIfG%3W>IdN315DskxqNBK<{~jv1LRC%e=OYhj>j!!^`W`9>~T2G@ea)F1j4ycVJ|a
zhk4!xFzBKvjWjq;p;zmuwAw+k(jo;x7BTGZkQS&fm<5~bkEpJ>7nb*pqu4zzqC?+R
zny#%ZT{>tsuC2)A@zurjYwcT%b$G|eRhZJcS7Rj>`%3to6)xl<JUxL$w`Dc_G+9oZ
z4%MbREd1{p{$)ryHke*Ok*XK{=5e?ze3eFBf^a-ZlXh5kChM5<RNorFw_4vK?FBno
z|9df#uud{uxzS4NqjuqAdjcihFv5WIE!=cg6=jK#?UDbK{{Mce>X?dKio3X;MhR3O
z7qgP3rCj0RI#QciM}s#+Lg!T(Q+01a)}vGSZEqpw%YJCN{2P@y|LEiAbfkvvqXqfm
ze&W4)_Eeihgo*yK+*#GxbJbq)t9FqXbvKA!yc*ACnvmJ8P@|&ZFJL7fgDDNFxc})J
zmwVS6eXb_ps*fEq?yJ$4w2yGku*SL*;_o|c$OSQ6=45Y%wFYBkoexWKGCG}l8_VO#
zqG@!o&<<N7)>DJ0D?ALNcyvF}Z=c?b*y=N~z)M4z_Q)3fV(Kd0lcOXxn`n(Kw>sE}
zzC<bQHxT+G6dAr%(EUD#-X`>Cxet2Kg0GwT!HO`h?l6MlYA#UrJrkN>?N4X+e~`o}
z8p2OgWX<L4(B0SwM=wSra@PeWO`Jms;{_kCzOyt^F%ZWeIKr=ZDdVG7(7}Z3n11sb
z*U>tI|BB=Jki1e1dG1dx(;M<Cl1XynwgS@$UQDZfoLE!OM!NbYhc%Tfgx|$`JZ5wP
zLNwzjW0yVb8*gEFdK-W6>?z9{b({)}0%-c7ZhYq9sS?@U<EYh?!J;{ccX`u=CW_}S
zh8NPBVXJw4W+rC6^1|#YFHFc*!a;>If;}mm9NudDd&d%Noj)8uA2vxM4h7;QFJ$XB
zq)<aYk;^iCj;cmI99v$7?<=;!{PcD*Ih2I?X5+c?ln~xy?lROE#qnKFn(0?(b9VOq
zb9S`9a7cRwVgFKF=BjsrCWrcC^X)9*-8LqZo|S@so<|S<Tw()8=itQYSR9oTuBD#7
zq;1nmSG=t-qKlVouZ<NdRaPPI@i`XIb01&$)QZNYpTP2%!IDR8CEqqld`GGZBvth@
zMLsA%u=X1$>30O%;LwX#&hL%gE^pv*;|6=b?4KljkT-98IfLmY{es`G_neK4mt8;9
z3pGAD+*&7Ic0<(%*P9l>Zs-bj<jfi<$tg(RJoBa%4fAN2K`c{#9)-8LoeEq|1?PL1
z>&W&zKZdHEMTmOSU7GzQ4FU0$f|<LO|D3oDcgKYyYUdJC47H;j2O??rF$XfcD(3Si
z=ac$@6v<R4!Lj*zin@+fC@4%_jOzSA>K!<nTn3b~LmIKN#P}mn$+$_^lneN6-Q~2<
zxQL?IE{b#LPv*BrLS=a_IkrxwdtV~SSiTwc`>QCXlVDMK8q<2wIjOyvh!8C!8fdJ;
z?hSv;e>&wOW3CT-*r}Op_9xKnx0UodBOT?NCW*7ee)7Mp!@DF8#<r1naHKvCnLX2J
z<gT6c;N5eH;T-X8=^akp3Uk<E!B9OopVR6$ZnV3HFL(8QhpNMSBy;vZ&HLfniaS0z
zNS027!^EZh<f;i6{aj>@NBN1ISqXZbeMF}Ucfh7y+}Wx^S*pDu;y-<%;=@&vn_UX|
zl1W9(#QQx(@)TT{xsUlxD?!3(!J|r5WhFs31<$AzZh7$}S6fRyKecFGr*Z@sgyP`N
ztB^%%qVoM4OmwWl@c3Hp@Z$wGS`C5OvdhR=u$B+5&Lx)?JDT0w0g{|>l*VV$?RhaQ
zbWbblPkF=s&UbD)SrK8eB9peHn4CjT3jfJ#TH2>Z)=_^Haqhxox%U)AFdLbE9g;cI
z5@_G&9BfgwLPnt}UL-vu4XbRXB3Vtjii&7-+K=VSZV6VsKSKP4i+N!e7CvY>U%udJ
z{wo8K-QFj>i51SW5#}-EXXqw5oVJYKmRbmJ<VNI$Enri2C1SehmzM`$rFk}OlIFmJ
zBrm%Dm)$p^Aty&N<yoZIV=keHxxcA%$XVt&oDl93LO(rQc-A3*o@8`F<aNahWrqQ#
z*n4B_m+5#h(~h}hS?SOFtpZuW7`n9K3tC<n!$xfpHHrH{U6UphBVOR|Qb+o?!V6_(
z&m<W+!3cS$E@mi)VUg~{w}tkgL6cu|x;zZ6!;W+JF*5GFz@7ElKZGAU<;%6den!!>
z|0IY0u9S`3Y)4<;g@Ak%;UrjkRUfr!%Aym9n&O6=e+yau*%GWY%|_9|IAk~IQy;x<
zQa#(gxM%%CGRDtd|JE#p0+pCb97r`J<2GCBcX%XiurZ?_0jF^)VJ_C+%wi|yTxjw7
zQxq0{hW=FNv7B2{vgW!lJMPM+N1x39SvUf2v+JpcM30{L4rUh{&$IbD?h<RW3z&ZU
z3zO-J{&>$@tl_>fzdBKdWbT5j6+`jR)Jrlf#hO${P38Z!ZO4g`y`{S9)}*KAKwtCj
zQ;68B%~pHBowZLO-Jp&4*3l-dOCoopFcQ~Y4DjrECh1S{BlqVsvE|w+=5G{3vt?Pd
z=Bq7v4ETokmwlM|gjiZMH3ljT;dp#tCV5WvVQ=?zLRQIgKJ=6vvd#<d!$e(t^y~$*
zpJQ1~SvjgK7h?y_;ZK&YVA;-wxVd=(DYw5u<%AU&7@;O+rT&N?876tcd@$|nY|Nf!
zg|LK4G*{$B)?FAx<3A+G?u(g^@dP7!)Tyh~z-J;NJ|*F6d;p!Fe}|2H-C0_zD7yan
z;czi5A>Gg@nW~;Tc61CD46kL#^-W<ut8cL*r#EA8(nCZWbt(|h7QU>vGrSL9l^ycy
zEPFG`kmvPLC@{3~!O@2ESmU@0PE#x><VKrpTh3J;e$x)g!y4FOI~V$B+?739uLloJ
zPjdbrMQ0vX<JX1ZB&j4xWK2SmBqT#;KP4n&O30kd^fM%6u834BP0~D4sZb>K?o}xv
zAxT0RQY11Yk@4H#|6NzpIq%tft>?LK_eG4?=c4$~3l`LMHRkC|#i__J1Zy>;r8a<F
z3Y$XGy)fa%<5;*hmyWfy^12EB81yxnTPO|4y|DH)8`O9d*4=ek*{~8^-4Vt*juF0_
z`z~;*O67N!WK(#vzby9DK<tRu$lq=6$JBL?Fq5ml*`iy`c%IOQ9D{DMcTels)1e-;
zQtl0kzxR`Uz(@3bTujFO8tBMOWh@unzB1>vEazqidSrQ9X8Wq1ET4A4_2q_edf3AE
z{19>2x$@?*RcmCqEEu2r1)*_760P{6D!9ZOuv4{}VQUn*&iqIA_xIA!t#No9dJelc
z=aLlj&n&@U05uISLXf!ctl#Izwy##ne{oM4A980>p??I$e7(-rrH!X@F|*eUnN2fU
z2fjyb25bWAa5QBS@@w6wZ=Vd(dMtc>FK*Ds%y_<N$zj~7ScI4yMnjs0QRAgy{K1Db
zatnV>Z>!DNp0kSV#--D=dtnVdeJXFhVQ{?IkN8MB#_HUCMFbUl&*N*nUT{-72f3JO
zY{-CNsL!gz<>BYWKE{`)IdDGB^gJ9N4kSM}iEoG~f?f71-lP34t-C2$K;^dBK^4M@
z_Jg_Q#p0{VDOxy2!i#P}P%^6!y~J=tJ!!!85npgfrII4s1uy7fI-YjRfV^}9eGdmi
zuSLnc>5m&WoV)@4t$`(~UXxv$J_~Ptayqi#5l^0SUg>%Q!{)4@?5WFWN_rH@Y3L$C
zwHpFg?Sqq3rL^Z%nedYtVuQ%CE~F?p>8~M8>;JHv%dvWsjiOWRN$whJ#F@EB@LY5A
z{6-32?w!r3(lFvPM?K+guE%5vPsBZHaSqy!4kX)IZcw~*iK-uY^2&;zEGjvhHDy$=
zozw5|>yG7Qa`_QOD!HP^=L;}VSw<&otNG0>j_9#lfq&Ef#~hv+BENMjxALrE$|h!F
z$6f)s5Nm!d@iY|rJm<f?6iDYkADTAUfMPv7Fld1e>^#aP@Bf}*y4G2G5u3}B4(>r<
z#7ZI$KMWmcgS!5EX!~w6I(YB8a7rl2t}OTr-R>LMY@R^+BCq>atv60Bi-*zK`It4i
zgSqzYQM^x2!GP!fkhWJdZ=C>ol~o{xB^QwHtqbU#bp~5jH#5iK-<fr+CzkASVs_PT
zC|}FS?WVz`TY<}IZ0uf`?Vm!&|B1hwxgG7D(w8T?`JvPLuk7%!)ihw=XjE$2NJ}zP
z@Zm)YekM;NgTAM*r!I?3*3FV$j5FY-3eV|Y$G3FiU>}ra7r~&vD%p*e@aO(q;s2dV
zo4c--?ce4jJb3mZFB*Z&w0`{2buTiz+`;@x)=^&mw2W;wjzr<_2%fwB3fnf*hus(5
zgU6zSUgJ9fCwyY?_OJnloJb<`11@AZr4B}8z3Kh#U}iNlj_Ri;^B+^6$Z`{Ps4Y~H
zdW&=K<4bXvl^%$dwV!yZ`C}ZmyTc?KWp?HIbt*o75c>n2NZ+dlnE9Ec-%nUTj0GOu
zki(zv$;hc*g|Nf!sJFQW8>d`adoZ8-8vbN83wvOSXOiG0`taL2?wFLjm<AViA=imp
zsBoD)=6qYt2How%3R_djD#HR-taTXS3R~~dg}gMAsBYjYtX}hrCpzbn$xFd5Q>Z{=
zcn0@eTp)V%*QH6~d6Sqe_+#&Xv3ZAj(89d8*c0-EbYB!=V1gqy?6jiRw^>~E#R&=)
z`Bh*4fygn8pfBbfA=hmNEO*$le(73tK3R>vH%G%kJsd-Aj49!jL`NLGQFOzMEBlA>
z?7A~7_HjDjzPeiap(Q%DeL}F(dmx=wd=1B%+c?tJjLGUR*~(3CnQB2tjDNPCntC2Y
z&fJ4YP<2Ms%1g|pX$gAI{(@J{_Vlb#SFreQpl+w|0o;1UKL5IiV3TJ4`e#>?r9Hzy
zKLZSZ+zZYX`^oFSoBYU)yZGB8Z=O8+9BrTSn1>%AZ1SH@eS3D{&Trq#GA35hdHK!s
zxjIt3pOqmg%4A>bpChc6=(S*7EfxJ)&7Lw8Y_6uGF=zOJn|<Ii+l|dUd5m^ed82=M
z1lzenm$I(B#_#scvZ3QtaC2A%9-ezGJXt1~a=wn+3<~A*lW$4sO8Klpu1ZQDzK3c*
z7tm(ok?@-Ii?nP`^4`8{skqUTYuorxar;O<V_N|N&c272dpX}wa#?b9>;%Pio%7?G
z=D}^`SvvI714EY2rdH>#{CmgwG%W8iEqA<%Fi%4a>pcVu<ockuRY&vj^9EAc=>%FZ
zqX>T&?n3P%6AIJJ&Fem<00TcINogKwuu4tkK_fr$)}pb946sGJ-ER7&`Aeq!y^YFE
zwdugy3)pRxN?C&Wxz5}W9#4)^;<qs<PYwfHE1w_JSE8mzaWaLnqiCJqi{{#$qtf{|
zS=i&(X8TtMNpIYW5aL_JUdP(ejD|W45nTQ9xks7Dqv5o6-vv}SMDU?@8Z_JC9cp6z
zvBEP2=(I^>1)|a4)>D$Z5D5LMe&p2SICZ?7Ltow=LD9aKaJtlhy@C2DeD#yB9cs!0
zVr<B9vT%E(m9XG!;n7`ilhsT;OF6?PvTh|tl<>iVHy;flY7R!wk?mM;^CqRFY0{tZ
za^?f8S`ZX4gkH}XPGt$Ahc3EF8|w09ollRz8FfR!85skuor~GNBX9Zn#5_LlVj4y2
zhVxN>-EnvG3uZTKA(C@S5!EPT3y-+--Il>Pv12RyEb>wl3R7`JeBFY*y^x&N0E?|J
z*lxkA>fPi{YKvZx(hv(-&Wqz@b*%s$mQ7@S=@$t1kF)X<M=-3fa{kO2H!(FYgI4~U
zD0tP)=<6;|WrxyHQI|v|MRoA>6I~mnR<qZ+`Lg3fJZaAFp;UV(0GVDPqFWw-Q~N7e
z$=^2KZ1R+!yJ3JZt9CeyECBD^*}ODA8s(*Ls6Ay4b#ZQ@VC^oLV+q<~QO6=$f0D_x
ze=O11L-vMl3BJI7*w1>+gO^y4%XD4Le0YZ?4%Z_U?aukXv?A%lvMwyIG?>`Pcm(q@
zzV>YxHjSTxGV^zArBficovjRok#BiKNFs8N2D9C6|L><gVmeuoOuZ!u3Yr`6`GPGL
zO#h8h7AaWyDuZ3$reJQKQ%;KJ3N&@S3yM07qP1t%(%ns~xl{gbY4lKQY^ti}=cnhP
z<m?o|lnT#tyikei#(VJUU&klDA4G8-k0M_$3o5&+n4kU;uJ!09n>!&JeIjo{u2(TH
z|KLpj`VM1FO8?QLwa@7Lqto~`Ms#lurXb(ijauw{V4M((CaeC)eQk{7)>7`<W=btL
z)-xBoQ&jEQ4LVL^p>eerl?)TR)}7HjTwmmKlSU#sTgiOI&oAWdn?bAI_rhq=qo_Pz
z&K9@}u0e)^`LxHK@@>5qGJQ=m77!AHPXWzPpBE+TnDY}l@6;(Z_BQt@nNP7Jj!}<&
zBz?M;4V|+uNpqVU-`Qg!yO?82rbZzclKqF>J!MZer-sVb)mb6wnIpVItnqAP9GUuw
zz3i6)WZ9o*&;9Qr@2&7D*=l3RDlar|48tk)HzM<0hqT{r)VXyix-MRY)Xb$Q+SR~!
z?*%QJvWS<pTF|~<#?-H(2%Yz=VX-;7@GKrddqyUa$DHTrUnkhsA&;Ofg<!N&J)g2h
z4`&|_qntP4XtbNiPYuf<!#Rbh>obY<zf}&4`WSlI@`@c;sYnHCTTsdVqo$|~`eQ_V
z(w`EU!kALQ#Vo~}Rwa_35P{j6_M*S_naUi`(YFr5i~KeT^IV#xyJ08cpHRqKyLN}Z
zL$F!PojtTuFq8_G-O9UGqKQ2VUeiOB(|lj}2Yj^=*{=R&$kuwr#+_V-lsVN1cp8b&
zS^F{DU?a|t3}Y9K70vZ#kEM{HSrjn57j*U(V?fepRQYDG?{`3jo%~6u%U|?W@dKOq
zhR1oGW*^Mg$|61YOBLtG($E2_`3v;#K*6I?dKeSO6@0^~rVHnGh30Vg{D)(sT9^?_
zLasPl<r`T;4iCh6yohUUjzZ&XKiT%xmh}E0kzY+KS+2T?ph$JR`xplKAqwWcXK(Z3
zQ^x$>+#z_^KR`CYu{TVAR${5!eSA|$U`2L4se|=01S@L7b)y-U4Gf~Lb?a%nrxs5v
z8c3tNzG8V>RbbbC4%hz(G;(%F-liCh2>**zZ6asx<$DFi>lPvHZY1Wk8bJQ_SZ=MC
zL7|mF@M~<8xsG*V8as=4`2`d5HT!~<l^toC<2ejns>R=tCe1Gt{W{f8tiS}+-BcU*
zyOv?Gn_wVKoz5m~%7a4x5Zdpx8o9apC~<1x9#`9Wx@$XgOWusd3n!!g%OPsFI*p}Y
z?J#mxrC@lvvy0t#Qs_QA1ow#GIz#_3%l6SUcltZ2v3VrrKK#gQdz_ao$;~C%p-V_M
z@TZcN6-fV~iEH;_sG#_w<TAPs-H-L;CixP3o7Tuz=k=yQ!Gk^$olf@@Q>fv~F7|iE
zAU5=tF5SL$0_y!%(fNvgRQXhJz->g&F0)j07ozF&l`9BZJ_;)7b7{sqH~bhR_+4??
z=(M+<A22V$h3>y$wW?6^e$fy2w*QBhx>|g(W=~o=<{D0oDMa2BWBziS_&;4p<#)Tt
z!S13xJ;|ED)|tEW=#_5;yXY+9+x=Kw@pb;Czu*t;5svJtH<D{(BkN#YMPD2O(PQ>f
z40D<f^MkiZUGEB4dUp--B`;vyTV!YD!`ZCmC;8?Y3ru<0gS<-Ps4FkR%ZsaMSy2f)
zJYqjJY`MbTi`Q4-lJ_jTtQ|#V)|7Qefi@n<k|qZ~;BUr{qF(FG=z|-u{kLF^pIJsb
zt{k9!V$adqI-H%A>qicX%Wx=q7kglR8C6!Ix71qAl9iS6U93`hWJE97tFcOmFMbH+
z+WY)!l`@;Ou8IFw8$&bjURLaAjpq@+uqN|A4A8!TA4_6nZE?9g+1($`^+gB|+5?Y-
zZOGZzfXtJzD6-na^et;x%6vbZ8)6`4gnP)w#1OuMQC_0gC<V?x5B+>$=AL9D&d>$C
za*{vN6@qAwrYc3Kn+VQnGS6F;1EWK^I4t(WBR3d8ua7!Cxp0vuuE}MthlQ6pe78*V
z?G*fMQ_N4!yTjL=a7TJk1dO{L;gjT(aVJw5GR;R=+jJUUaknN7SrE#yYYM2LcmdZB
zXqI)}d!6^Yx(S^-nvlY|wanGB1_4qK=`^l|?$ZwDnsqr+T)ID9$<(Ect`#&dWpiH6
zXD2>!{tEaeOZZWF-%K&u37!fMsP5KQdYW&9MDwGv-+?!!bwgM4rIRA_d?n$2POzkV
z6KA8tl0ZBR*@!!L^GL&RE_P2JLPc7(m=rq~A3uHOi#AU{QqyNd6lTf_c7CA-R=^`X
z*U*xFrBFD&f^CcrK!49Aw5`)fiZ3qWId%rrV*G%I&hq7VYJ=%+_xJ4ms%wH_8w<aK
za%8e{8SCxzm4>A^pkSy5{g`ToDQg8s-OwF=KhLs*7Ky^aRwR8;c>`)&hq+(MsN2U)
zu%4KO0?lN!XM7S~;VE=s_c2m(x(Wr;B6N8$8XLbG^6{m~u-lyhjWO?8eEvu(i_#Rl
z*+*<{zN&f4uEY3i*`0zz#__|voJD$#2N^oS?m}OL_Nqllt_kJdDaE?gFWJwgkFqIz
zJ~z4%g%qnKx_CScgCfqMa@P%5`qkp(+ncDHa~1xguRG(@NOscr5*iw}vUrDJ>iVIK
z-#Va$3oECKjN2CM9+fN^`1@kp%K#jhx(@SAcTt6-4WgCj<JK&ZpBLHUlcM|jE8(-G
z@WG!>JliV#72xj%4W?J*!F)TuhW;ZNzcxU;7Mm|psb4#Ox}7ECa~*i($5@0lDCA%M
zWyuab61?X6Qh4|IkE|mS_z=?_Jb#xauQ(bfG8la^pS2+?MU|2t|0ZpX5M)e=#^H!~
z*6pf2RxFr81$#zQ`@KGBd=!%Bv2-h0&bdb!1HFWY_6f%J)`iX=Cwf(Q9Yezv*)HEY
z7}lJ|Uy<|PljKGl>~v7Ht{e598q0cCNK(UqQ`EYB67z5SN)NU$Qul4=JBG<ndE^hQ
z%+vXpMo^r|Qweu9dCHK>^da{YIxYysk2xdJ_D_bUdlRstD36Rb1|g<Bh<pz(qBP@a
zvimJ=+<ApRUA!yD9*Fsx`hjn-d8$AaUOBYb`2u?>eWaoPvUuOK9rN#<mqX$KPxNgz
zz}n$rS16r=*~|x6uAj!@bQYqmgB?c9XktGyYB6=jU);}egQ{5r`F8O}q*elpnXuLD
zmR>4Vi~d;rwJ|h!S0K$P4S;jR1Q;dV=M|cjcxUT_v;NxvMIRiv&>>$oZWAW<_<{99
zJTW5YCcPZBn}*D~j#F!H)5g#fRFNZiA3roCo1<ZrtSj*Vx8t-i`8E4wx0%B3=bPz&
zJpsc90hspp4s^v__?^osj9zn8cCGRc>V}=84_7xr_FIQ6tgoV}=q7IEkEOc&diJw-
zIX1`{<NM*;EV0xWAB%RNOy7g6X_s^Dz!*AUpM}Js>xH{}4ZRo{4%6-{S(`-*n^!kG
zuj%C_XfMlR>MFiyW1(dJ=`3$lu*b3hAEdc#$85{>u+aK2(a53~^w%vX=Vgld^0l#$
zA9EhjH!ndo!W*~Yk4w!%q58fS+S>*1Zf>jDuuccbVwCM9xN4Exfo^6C&d!2d%Sw?A
zkuzU<eiqZYoq=$-DmL?mE#!9vkkQg|9CiE2*6F&Vr~GYf9JqtEX3i$-O$($g#~kn@
z_KwJNofDacKBDt58pk@Mpojh>d>IxgSmj!jacLBLxZ?*>9)4w`st@DC*Ekl{OE@<V
zTt-WR;9M=7OM&ZyP#HIyE&b|34hN^h^xXq~e3T4^Z^iHPt`Jj$SF?p5gUIXiZaC$M
zJLaSF@c6cpEUbR#Mb7ah!>G3~w{L;<;XWj7@k8XCWXe*i!vn=Ue(dH++<F;{==-m5
z{?JKYb+HqsdS*ba%Tc<xu^iIg-NK>n3vIo((6xWSG}ry+*~db0;X`kFwmXl4cO)Vu
zXeT?9r2^B0Oy)XR%x9u%5cXmzpTG7rciHfRoW2MTm+@U1K0ZS@@4{i>X+bju11Ycb
z5oWs99;IEz(bb;SJnfSwS!%r&?zLnxme1!;59+bYfth))UF7l?Z^&f5pPbHfRyE@-
zcFE-QNq7P>29WclRcMg}`@mu<R9Xpv<KuDg#SU!THkKK$FemyHfO-2nkYVSMtY-Hy
zES{0eR;RvYCd;GQ8M34F>5Y6<|NXSHlZ;ZE^KjrlKOXOMl|n-Yqo6~V{3lwfyylfB
zo>gkokl?i>yIaOP)@m{NyFP4h+-z)K_71ikw8%N2mZ4VzYWD_$1%1ToVQ2V>4gb*r
zpa00gPcA=D>lg2`uVa4kg+Qj8nk4NBuf;Lx2vfTK0~=M0sl3ZWjEIlLhEdM=*k(@r
zyCazjwyOM(2e4fmgHENV;XgP5kFPyP+Sgw!b@M5-=#^2XV3`^@?ZuqE51G5JqwuPE
z$QI8W#a0fyfQ;`sV#j`(>d#$gHs5Nc$!d#O&qEm)Jm3&UY|8}eb_<OmB}f(D<N7su
zSbp7uS2zALl~X08_F_5R99}`+7F~pzwI{b7lSLc7;_>KpEqs3uMb*k##1xBgRdfcW
zj-JdpaI<7!I178;<)ZYf=tM0)i$iq`txwZgp`kz5@99BhZm;Ox+s?89zw-FQb;5h0
zJ)J^do|08oy`tjrHBwZ~6HK-`M;nvkx$)i4X3qPca+6M@rACJ!c2K?&n_oWUj|Y2k
zy_Y8Rj{9I*u@6>-#o~eGYTiG<l6u=!Bi~IC&VB#ja`;zttJ0$D4Pu6JL);TCFQUql
zHa4>J97L~f=ISRdlk5J;vTZpH{DD>`T@RVh96y``AFmH1`+jKeY{k0Td-GL_nQTv$
zT>gnsm#JLG0TWhb^5juwczi2^{_fDjlyD8MxYG`&zTSl5i_^Gwt(=lKpXVI~Grht%
zm3o=Ez)tMmV<tqAtmi7eD6x$-TjubR86U9YgoNg}5IoU5$Vyj^q~Bc+v&fA{5Y^T!
zTyN*lrFWP7^P9RsyI=%D?|(-}k3?*_5X5HtDbNqQQa<n3eg0rbprn`<&D+P+l2@1O
zO#b#b%-ny3!bA=vVS5#2WK_~a8#x$`y#`r=35H*Dp*hQTVdJjJm})r%JEpIK$%`Y%
zN;n3qRx5lSG8-#JPJ3l&EjoTCS;L*%&?wNOoKA~rQJW2vkNC5Xeh=x*i=kLE_X7RB
zB$&kugiFt-ktIJIOJ^oM!If`k==NiGrnl3BHcZ})ErIzg%(NP>%Y5;Cst=Y#o@HEP
z6KMu!(}!+{=*N`)w9ax3e|e^!HrDpQ*5GJ#(7(=FW5c1?DC2UI>UepK5<c2~MQQKV
z>{xq0sEPhnNu@U19`2OaU{}m#)vDCV+y~hyVWd|&j-QDZ&#j)4V7abmia)ejP1aRD
zeNiAz?W%x_Ydbb~InFm~+!5Jik!Om^M7MFB*^N-~jF0O`V@{P~&SfLXy5A9+_<k?_
zwW^{2!o*A=Uk3+1m4cd8c#>@x76wjWWkY?jH^~4kDy?ishCU`_CXrlx7@BI+;rvY(
zYkT@*Ws94vOPV^CMuLv$CJFYG6^^G4fT@iqJCT$Ng_unI9a0995yE%2FPddKWpU|q
ze~fE50h8PeW^c2d<RwFF>ZFdyAwi^fGfE1qxx_rWt-$l!jnE3JfS+|2s4c$3t;UwJ
z%*Jrkj2l5Yx5hwCN0AmETS!|~#cZR=B(L&!CF>X8h7zd*1z)*_b^0=7e@y3&-&c!{
zdl$Cc`7qA-&Y<kbQ}kd>C(MpHghHn-MEsFVMeu2+YAaA__8i1o%)lWHM+A!g@~+vv
z*wdp4OncNIs_$HhhBfu5X?dTAE#0wl++`-$#~lM#Bp_COI9Yb}=h4aYC_(iClE$th
zNlhCXru`|}@E6K5uAql*Z=@+D<NJO~_V-O9Uo}01o{V{jQ!4SyMV2mW?&vMe3p|bd
z-@#bxB0h(Vf8oQgNSf^>c+~OvG`Ky6>fBZ#_~llaG;lp{`~Dx+#&pWBn_w?H^g4iz
z-P?h#Y`Te_CT?i@&r^11TCt>O_n!asR+e-o2hfU%-rTIDD{OW~V9?TCSof}t{;XL`
zFZXLw>&`qV&vM7bs6z-{|D7z4C&J=Z4vutE%y(m1yyunzS|->Yo{Q(B!)^oUr}?oj
zs*z}3u#tPT%je6Fc4b3-7t`V4UCF5b4Lo{Qjo-y0f93O@buw|l8^vbYe*U<WGODxr
z#>At7Pvl33H*G}Vk^oZoUm~@Sk%MA<DSO)75tDu^a!JXAZyV%CTZA`{Z9R@3PXxcr
z;3{e^er7itjv)2TVK~dFvZuQfrA6~U@cxfpaI>xvRIlnO<=%S0s~fFpXXbkBa&_ms
zKIPJl4<(q>*u*VY*U<~b3Ya`{#lgdxh&6eFP~RPB`s>L?csxLWMK$d?l*W6!XyHm{
z-4Imyoymp_CplRhGuUHI6<Q}@Rr6TzUo-G2*@2Aq4JMbJx_IpvivDhSw5jSlTW=JK
z!_VxYb#*OGIo}D}6OO=HF`Ij&WMco8aI#HZh;4b9s2WrO1#u229yy=hxJxh;e9863
z%lSQ7*ZgyfbTD<od}v$>#3I|<)X6yvNBuv*r9Ov@)+^%gAS<Rbc^B-ScQd~{Z3JmA
z`o@}$r|^ZLr=iid7p=RMVZAH@zRP`B*fg=H-{XU+^SD{u-ER5YtJC1R?x@VP+afq*
zdQxh_3AjbPXC<;U$xrm=p2?piebaEtZIPoTd1t6BIuAd_+sOvZJqOLGM3fZW<`24C
zvwO4M=<3(GuvvGV!j@dX5QhNT74wC^dF+L=Q?Ihk3Chg*csyIW;|(<4UZSEG7FhLg
z1KoA}!hRg{M{~+}@@a|`&z&;nw=e{z;`95)Fn^SGRHFB<MTck1LJBk(NP3;kXrIPU
zTo<|Bvw<P-%6@^W9euG$FAV9H8=>s>mPrQt5qswjzSp0kjw9R=Tp{@I@)BFoHGw>4
zzmg?*t><M0nmDt?A4z@mdE1r^>}gIeB5#_CJX{8Qy?ZT`4opFw8pBMT-k3GDI~MoJ
zM!mrTT(gR&GOaOmd_f@7-cf+a4#%jVV|Z@lXA>Tvo{o@(oy;d`S>oN2Jd9bvrIx>u
za4lJn0@K&HSl*8NN6ye0rf9zA(QQe6+Y&^MKZ#QdbNDmw(>z~y0(zfuq`65M*fh@@
zZ&r-Oue=E~_2vL<d_9%DJM70l%pXaG?cTI}rXh}g>SVrscq*xmS;i!7W41w2N4EU#
zZT9iJKVKa5TsWUEV?f&qF(dY(2!0&{x+vms%U`zb#dcQRZp9UY^TawQmAM)vVYt#%
zl%2Jvf-{PeMbkjDBwKsw@54}vU7CXlb7JWA=GU<4Jq~|zJlXpAO=Ov}hS}}bBln7n
za1-)xBUcYdd%hy~+!U(lwvRqOO`y$J&mp%*lk_M_uzR|EAfLI6?AHaeZY@3Oz}3;P
zm6G|!(TUh`N}MHcRr8_2eaUusJ+fyUV9NW}(e0{!lIp<cnEoyp=8?BZ=b_l!YL??~
z-8@u#F6U(;A7HdM1J6cnfLu{E#t&=}o&hteKYxV>Tg;~!l>^B1WwLarL}KH<?Bq*N
zNAqLn=g@x()Wyu}77I*11C>=72snKhnWx|IZHWizTEZ0JNa}@2I>%{}ybYxc3zO1n
z<mqR_D>_%B!>w0U<n=#L4dvZ_6rs3;D=!^HZ(1Ib?!R34szkx({u^enu94sUwTDhw
zoG0a++wAdi!8Q{N?dtRp{5LpRVn0>rUqWxzHKQX=s6??k^$Zywokm|e++YeD&ha=|
ztt_dr0O|^Zq-Q)w^wcUSYgPz7eC|R&pLRA^^)28JXU;`wwKojT4aVSHFUScm#L+Fl
z%TOQLhzCb`ulhE@&hzF?7ZhoWVGUioJD3hNmaz4+m%|}4S@i8=kfeJG+x)U9>E2F^
zt2-$=8~!M2?F6IpNcKzT0aG~Km!>Rkp+%o-Wae()$h7A^X4E5^E*5)1ZO1d-;aDTm
zcAUUMkyQKvNUulb@dw)vqdh_5ujXyT%;?F~y<ruOZTBSGPy1Ox=Ui&+Hyp0hBr41k
z9Wy!M9y8p@PM^I&YaV*h+VCdseAEXehJnocL9sOQlM7ArAI^MtU8Uq9(U|O00FS;K
z@KP^LaOu~g(<deK&a1khOYCK=A7=r%>Xpp1m&my6YL;%OJ*5Ml|5#+rcK%mzQCBBs
zF!`gYl7pcG1sd(+scWkE;_G*5M161U+z?M~!sYo>Wjjr|pDdoU>5xv`kj6wbAm-{|
zwsJ=X(|dk`_fe2DZ@C&mU&MV!|LQkxwrZcGytE9ndnECFrhh3`vm1r3u1D}GA8Fd_
zp*SBRI8xtr&6I>2)W*Yr_~l`6$(}46X0ym|PBt<pTTqSB2YAUU_-^I3_%5=C&lld{
zs-w!;-)}Z7s7EM1Em4PKVkMi~s~R8DHzRa%BC21PV^w1z^zS_&x#E>HbK)#2P`xMo
zD2%ob&EeWV5(Q_@lFQjB=jfi>MpYkXVDs7+B4c!bG|V(GFf$ABOQR^|LN9!?GQi~d
zQ8=1)AAX1WQ|yaqIO;3G)+h^Uxq382Wjm`3Dulvxk`}yNK(l@=C!MMfX8w!k!1#wP
zTRGzew?ExQ%6aoj<bHzSB(f_e?q2wLAsAgMI#Bzu5jcEg5v*_PQ0cHOl2X6<@LIBn
z^|k#+ikpmiX}|;0yWh%u?@vX2mlMd;x`Z#bAv9OHi9PBQ&#e76@B=^dF|;yXytgeW
z!pj^qvYHG#71OVG3z(iqykHRzg3*ism}`dP-TQVLC*Cssz5imnV-iMaRbcjyX{_yi
zH}YCO8{L-7f~!J2TC!9jc^#8RXFgy{d~9gI*x3~P_z5;E<#D&X3whzqQ+e+jWi%{j
zE%dfcW(wcck+|lWG;(1w6#NzQz1JLM*<*?*dczG_;-vss_buiF2M)%FyljlKc}Nvc
ze4($U%KX_9^s~>yAa74@bMd{bG_!y`7=Kb4w@{h=j+llqHe<=*xGwC5l(If%QFJdg
z0;%Uc@u~47{Wi#z#@;KWLt{lJ)Tn`SD*KYPrwP7~5e%|RZ}|j!cNFgG1&<<q+~^d*
zrjM}0kK6gYc(n<AZSQ1GqkQp4J`ddvxwFiCGxXm36ppXg$eu_3k*dZzNj2U(sQ8%&
zIjabtw6&P4Z23fQWi}Y=x|9~C93_RC9k{V<5Arr|reMb={%Wx*dCgNaU!;8*<My1#
z1Tk}8Yq?jpQSB)@M?9nBk&!S-@~0=Gk5h_DqnT{qWj21V8w>s1ms#}h55s6ewf6{d
zPbh_oNq@4pZ)MG?H$XSykfSCt!c&Xb*g3*wGU^HiY5E}`@eR{_?nM&>i}&LecYg1r
zaKjtw0jZk>i!qgJv^=MQ{vCMZav3JaD)EM1$LWm7fz}+Kg`#9RbAO*F+@vjl0{*O*
z#{Djn4ZZu2-{E^PI_f7Ic_9^E=L6`LsXN!n3Zmo{KDb=70cwgxEM}}Ou5Zhcr736f
zr3NXK_@gguhU-9mOEirQUr7m{wqxmPIow(LmW7GmIeF-3GY4&gYFid|Cw_xNiFe)x
ztB1mS-^3I1c0%vlSQ@1FSQfN7nko<ZqWjuUG`#O@X7wVL4~{$rC&L`Nu}MC^)7TAs
z=$9f{%^-Vtj16M$kt$fZ^%$Mb`VR%{ICibQ$F82WMGebAkoh?HpAW*SHg)>yXh=If
z2h+k~#ezHa&1}fX$ME|0TGlI6`0%=wA$46Sc0V2s*|B-}xZ)HHhl?y1|4uf|@uDa0
zjmcxmvAStL{ns&oOW{%I+D(~8<j%q1sZFeCbULlBJ&d0RUZQMQC^?0k$M9e3v`H}2
zO&@Nip@v695BM0)haW;o+hzXQ-9wr;<tjc}?SjwZV)X0NN$ef&pzlKAXQ?*9sM>PL
zGt?92{eH;Y3TIPKmvAKg@uV-Crr^pI(WR;S&av4Oh~6sp8*cRZOb2rZwLk22Px<^Y
z)ATTCTr2N7cZ0OUX8<`p3c|BrebLJ#Ty(z09DPkYH&{?0+(xg+r8(D3DYhe9JIjX-
zj;zA76K(WYqh3~!>7BRa>KRJz>p&f6ou&!j4{_b8Vf4E!G4Er34x&;$VKQhqoiR9x
z<aRS=x>dN};?ybDUD-UuI*q*N1ybp&S(v+~oZ4?I(qj!5e9Lh~$0G{Z6Pd;Acf?@(
zE|xbi(+gH=AryLc4YOGm&eBHaQ=p0ldiU-`2GfgW?k_8`TiFuzht8Q{Xfv6XrXtMt
zz4RvcCA??2@{CPq@n4NSqD(rY@XTd6n<S&>$ve0;LSp^n^e}za8Z2#j!!><Ip!b@I
z2)wz0dP@r=-Gf!+yyhV47uCW#+MX>hno99Y@AKsC9=Pk&47XDO(&$YI6!LQtyCR-#
zh4S}tA<YY4gQmmv&jz+DBpLFK$rP*ff*$r0{<rp*xLEoz&pNjR{bd;#lW&f*|5A|I
z`6D8iM)3<y{(NcPD6F2^&a(ZAxV-C8<gI=UZ-alrb$Nn4RnjL(<2IR^EJUu64*#uF
z3kREgnXdIzs3oN0Q$ik$Hx4wj{@o~T^1aD?GkP(Vwszbc5+^%n6OFX^NwSyMvRKq2
zeOw8PF^f;@lyApHe_&=0KKafIS>2yPv$Kjxl9r|#EUh1)N6bF#+b;*l)(v8h$??}N
z2(K;tp#7;(R_JemcfG!|Ltp1ZQFAc)e3?PpXBttYd5ly)cn<vze+lO|%c$VLB-$gM
zUAG*TNK+=CNAA=k=&c*eCeOG-p+EZJSj{c;Zk~@yT7*NsT}XaF3j3Tb7?FLS;==aF
z!r5dZX%uM7oVF#<+QNm1O4>u8&mA*U-?a%ty87YgoDeb-`x%#7Q~qsr9t*RJ!b^uT
zzPfBO=Ik3s^!br&Y)3z8clP48uQuV%_ipHI-<7OeBdL#PH6CFmUI<>;mY$>7S=H|d
zHT9&iw@jI-q$Mfckf)}v_ULtBJ>T4UJ=3cF1U0=)cxgA2yxOD8MvQ2XrH77^{F1hk
zqhS{)@;<a_s1dAs3?_}&Q5bblo>hwesIgTS^O8Of(d+aBuJh83+L8+RFwyD#y3_}=
zL{>|^dnB~}bVaV-D41t2=C6mk!Z33*yHfuc(IJ*pqOlcP)^SMRYJdm73&9G-{!+V`
zLUYE7oa<vYEORLAyDt`V#*eJ8dX>yo-<5P7X2H_(J)e?cjTLX3S;rlpSTDO!*!7x2
zYwOO#%Q=U7m^#q);!M`E=p2o(6Wx*;bGFIIoaP4Ri#zdutS!)qUp_%x=l4TeT{jW-
zl}BL_a9%d~Ng)OVbu>R<n?%ee9#5JaaUtDUxM5syzOSnJ-i8kR%3$H1^IX6$PUwJX
zlWx-VKiLRy-O0}X{fD26GErl@2A8bPU_xtealdWG<0J9dlzW#py=vouPbA9Qk%wc;
z20$~ki7&ptPdM=u7!7b@tA>=qBGCwg#SS-4zlqKK+reD<>;{(N^odNq_(9D{j!t@X
z<bO{oQrYV5$m-W2zcMBXi*?iJbeu1>wtwSCH0!1LL)MF#<49CEE90|u67n*4(&-r~
zh;-7!ka<;fT;~K7V_ML?<%d+8R|Pla#o{x(H+-&DFy9hW8vipDq0X^L_U}$%7615}
z{`atTWdckpbm-s4dvqrKC><NwS2i#HICJ%~$AZo7=w_TmgS1b;XP6PC=lckDXcQLO
zS|j$D4$bT|8R{#>v5&Rg@UhGh8iNB->(ZO%ol_QB?Kp%E4T9RPEUGj<L?zbO=!ec~
zYR=Ndv?rS|{L@@=nsJjNS0<yE;0-4X+>TibS0agA$*$8GDcvcGV!Uq(fAA_)Z(4z{
zk)6!HU3P+t)o9YJNWj|M75qTG845gNv2)0MkcR@cR1(PrJrVnaWW0&`i?vqeWE6ND
zIv*otyzw!1n<_(NxeiXX58yR7ys_`Ox77I68s=r*q_pk;1<$pmd->kzP`i&k%gtvm
z_CF9H?aes6^aBdQkCE-8TL|g(2%iSe!2s3&@a&ZnK4k30g6&I4Y48_*rt4WsGV5So
zKXn5;I$|PaJTan5S5@;$%TT%+aS!txu8^MPW+`u2D#bq>i|0WjspiRi`knC<dz<&k
zN<L*uPfNbTxR*FL$onvFt0?v(If9g?W#Yw&Tv^88*+|swO#?rlrvLoam{;~!Sz6vN
zdho}V>-rs`M4Nr&Srx^~7DTYlon0xR`X%cX@>jOZG7z^aLfOyh8`v@Ha1t;%>{eUL
zwv_Ip%#@k%>*fuMZevkD^)Yp-31IQ&p=ecpiQ!!qAW<+`5|j?&AY#byU?eho6_SE(
z1TF5+NHg=gn0u`5g<#bpE{Tlr!zuA7@!dtHSNEg)B6Y!5`DG>-uZY|q9GQ6$2w1aB
zn!RotdRz}9#l?}#*!dzG(=?yCbXKSLOMLljH+Ndr_MEuoPG}grBI=4VEqUIZlnY!?
z_T#j0W*?z)eL1jF7n)k)As7co@MxGP{nkvt)$M(0+>TXL@j4buB1A^vEl1tlJmEH%
zGnb}%uuf{e7&RqKbSV2%ZKJr%w3nE@*VxST76h{`^+eW-ys&fD1oq~kHo1Ru$H)u&
zx%N+OT9p2VhA9mvgWgpb;2+8St`3!~cU}|wgjCA*xWGECJdRV%eh4kE;PXWGOzpc9
z6<8HcGVdOTQ_&tgAR`T%x6Y@1<I-u$8F{GR(J;3^6h*0?9G|8*qh>=OvY+Nqr<N(m
z$U7jr{_hojtSAOqKa*|PT}|oV$5Q;?leGP)FX*ZU{_8WCmJex`!sj)S<@|0Kpf-;J
zhrOdgf{}SrxjPNF2!XouT!a)5l3SuMDB(7mR3mxX@GAHpxlHoEcH_T-QZ$Wz!BlMH
znN{c-3ZA@zV&}$5TL*}<;EPuN_+|=a9lyo|I$PH7b*11&bL`vn2UDkb(86PiDD{Y=
zPCFmbiJUmPIC2RoXj{R1LNo%!T(VgCF0{v9;+HRdK<{Oym|EhAo9nL8gRMm{4T~iQ
z_i(gt>_;h!9N4QV8}V~6r#A%|(kO8*FwoORrs+0(e)^Uk91;5mt5!Cz`!tZ{2;>A#
z6U=Hive#*);(vGHqgF^=PKn)`LlV2?7>$&q1pHe%5i>M<W8}UwFr4+7f0_1@6)GIY
zy`;yIs-GvmA8V0@_zz`gkLQz|djvh*bAWfZ45uXzJ)ka^iPEXxF?d`cG9zX(`+gdb
z4^qbg7ZX<h+LIZr90g^;fST7M3VI>^QC9F-dgWRSlWY@AJCeX1mPwT5dYl|pB5CLA
zM{x6cCzZgCSL**_x5abe>5`R_<F{VCSK}p|bqk}5RR_c@xEfb^EG4|Xiar`c&|0*J
z=vNf%=OmhaxiOOI1?6&0^`n><_Kzv$#b9XmUq+8w_{pI!MecPMWnAuTzWz=pen;mn
zRQ1~>%e-I~8nqO@#~0FhmHGH-Uo4z{o$_zET@b9LN4!cW4DaXr<Hr|gQr%!m-wg6)
zQOdRGxBVV&6u8sXU;D_vC6>H5HIP#$RaW5{0k!25x#7NJEbGoUcIbl&O7aYNb(*2H
z)Lx05?k8sE6X(;4i&w}^F=S%TONC@voQ&g|A+Y$p9_M_;`QBq9v+)^BeDphvJ$(rW
zx5n^=VJ^g4&SBBn#qeJ>0n7gQ;^U}x^tgH#8H#V&v{8|?tXzk$c4&hA_zq_19Kalt
z8{q9Nyjc^6(Ik=6^VoAza-AI`(_7~xJ7HjiMaRbQmHrRd0E2K;-TOi}Z$zWU*n`Z2
zOZlrB@t^2fnD;1g6^`4_$CzPZTtDBQw6#{C!+m)&9C?_RIUW(dncMhwP2>*b-pPuw
z3FD(puwU@>&Z+hx^$yRma9JsTYBz~x*o(gH#hdhRL<}1~e=)R+^$@LKN#Qb0y4hz4
z^-k+b6RL)wyw(cQ%_)#`Oe2+fr|>3XCiNcmolTq(gU5j{;It)}^vB0=rCH}mao&6s
z^?g9UhdjjmA8~Lv(w9AH8iSqtpK%|D=WO2bM7r{IA|=JX#>r2CxUC#VjSExYr)Y;?
z=f@(BrsB(+5HPRK6cMZkuS++O{$T@6J39@#7w+KBXJTOVxPi~=Hj!VRdK1I84Mx3o
zJSFFa&~d+~(#l=G*o+m^`LOl(Sk(|6Tv}Sd)7E*y@}4*O-d_Y;$5J*X*bi?zi95@E
z53&f8qh#%esPESulA#Xo<^Kwqk9$#1$Mw+ocSGdYn)yUERpEr0&QfYqktp(^rV67m
zef4i~{_BW%nTom8Bc5!_$I<tpZ?Mp0B3|}ffUx9iG<Wp|`qESi#f=T1Aurg*EsOIe
zS?S};-z%aMQO%V4oM*c4kHIEv4by0-r0T`{m~+!QR(CWD#ezh&YKj|WP1{QcdgY-l
zOY9VL6iC%QR@^IhN?PU%DZsQ8m)knBbt#(S?0*v7=5ORzUOq>^Ks}rb84kDTD-)yN
zPeXUdBEI0sO>*3JRTgmFjam=L=;q`i^tJN94Cagp%Z}pL{7_cAHvk*WjPd#6C0rl4
ziE<ackmac@qb>KtX=~&V_IU7U%;<ZC#ic$(fZQ&sYkLdF*=2l=-*m=@X_G};It!0B
zk}hkk<m1g1X<GdXl;*y|F6U%aXqh9!HVW16;-sm0!`bH3F>K$UB9^}U7$sh@WGX7r
zOhfS*t+dzz`S+eQVWtfdEY`9Y?wfdKMJBw*<{{H|I?a9ei;ngeZb<Pnf3pdq`p>_m
zF@bxT?v=Gr7`RPVrC*4ZUVeD$(w}d4ScT5{&bYX%66cfxP<y$Jry8{&d9mOszEw3J
z?-DM_m1|8Zuomo$6n*r46vNhG33Ylo4mjq(G)$LCnepLt%~pf{9*pGKGos+&bp;R7
zY|yCr1b^cM@7`bpMdmw*eeN#WrWwJvK1?Ok>;tTJ%@cMpC>L^<gJee=#K6cc9`e`4
zPM}~NUd^4sY>Rf|(S&EbIF^xIpre`E%cIaAwiwrK<nl*8@PJ*~9no>_NuSNGq2fw0
zc659N1!XzzJ$EB=lWOSa{MV>bJVZ<Xm7~M#LL@Xxus(7K`sG<N*WV^)GO=%8bn7~7
zlIPO;!VpZ*SS>4eTPi{hYV>^J71&=hrjDO(Nh4OBrb{onv-M|&v(M+}lEKjky7Bcb
zat6EMY1db%`jtubrK)uA)^++EoRMevs0AI|53_Y&=D=o*p3HZ|8@S#GM~um5{{7iN
z=zdE;(Znmvcw8KPBjM*z_QXE%o?uRnf&<@znyGFg!x#XSE6eC%=1xAK{FwMW61hg#
zw-~=hoRex+l6|4G$WU5AF)o{Q)L)=rTPQbl8AiWKH^Ok;Xy)L4H&6GNnA<74(2{w@
zcoVaiycH&6gK`!O|D>XXEfO7?YqY#m5QRC{@;lQxbQ)fv#}qv}G5QkS_OfNa2Ufx|
zMlj~pZZHSUu5@w#FInBFJ1pZ{1daT?n?HK4gC#d(aWTzT<js<xd-EP?d^*8q%9laa
zNnIk>9_G*5T*>SDFboyB(>22v!|U2U>E5aSWYK<PQvT~z^zp??EbaVQ_V~qIp4zel
z$s-POx1p(+m1cqQsv(l@$4tC3lFMIoD30dsEJxk5qx@4t6?+H=+^RCgTayylY-(Wr
ze|ln}(>1^#4U1iFNCrV_>_A^HTChnl+e{t{UbY7{r1*&L?p!Ie?lOh#&>^G3PISPj
zkl!T*^8@5X1(TF$)$|J(Gxj5kH>hP}ZG5TXR0JmdDG>MX9qim<eIzbc%+HTnh>cy`
z_~+Irc$F+c$?)OWdOBBT)6pC|e$R$Ot)<L1cmr9M1aSw=yEOIpB}wD*W75@h#E&g!
zn6K|!-uf~GuCIN_t+xle*!46i)Sg2|N)c(Q3?RF^ZuID^8uq$cpu%G@EN}f`?r|N+
zr6PxlwR1%mM=)pSG-1(>J@g~B5zc+)iJpD9)G|}d`~<c!=U)SJxs{1!J;O=wOgUnU
z66u%U99;UKnt%Mh33=ZMq60y8(2CiEpqwA{Twwtk@?Wy7`C)AH%*izPX)GOgN@Vla
z8{?5qI!+5N$mopgJpO7iK9n}IBPVj1v-caxA{~>~Yc)~P@D@1Vap$)#*pkE7We8u;
z!F<w>J$Wl4N?78%ee`vqgp?^9kHjv=_3&a;JiADNQ+l9ZTL<&1Q{!>FN%#&2i5<hY
z7v#Dl1CuPh$l7ikYVMrJ{?Wf>6<e*~RJxFziad@9T1OB)>9?%rn0URL++td3+Ay{i
z&%?nz5h>U*TA@2|%X<_wc6K1g4$tt*zKptH1}eLq#!bsxJmS0(etziA4dQ!~k5Uy5
zJQ5lE<MT)+UenjMCG&`rDWp5tjksqCPHlN4_7LAlWyVSj_%>Z|QRJAlg?K&nJB<5l
z)sVhYm0}kJAZmXOn`zt;{XeH-s{R$J{HPm0I(!|*cG09Ud@v486z5NaK&*e}fu{HS
zXvMu+m|SV)qIynOhRB&~r}x66Y6-35KJ(y$bhd7|v*12vQ?yMSIV?(KUjGHLrW;|X
zc{qUg&dw&4_kJw9AqV$a5!O_jBhD-W4ztzaabY)XHIu35Lc)CY56svxh0ch3OqAhu
zsJ%<&dY;15+R&ME-!VKb%$<&I?m)U0_hiPo(=fVhluXWEX+l^|1ly`UmG{rKmNJC*
zwlU=p>}w)P86|w$r*>YwUl~_+Tt&BTeTaUJVL8f6u)S@a*}OP?K6d<3#%Egb?>9Ut
zaOiLPw?;S!o9#&6z!8zduhZ)BUc$HNhIv87q`6$19JT7O>PSDj`&jtsLjQ{XdMr~e
zx(9=X?l?9yJ<riz4bqA>#Fng(qI7>^zK^41u6z{98&&ds2MJD~c<z{WT|?pD?D>u#
zK7joVO6$0qJ)B&I&D~FlOn((NKha<+1~V|~%0)g{^(uxm9A^$%lf}=_!dmam!|Rw_
zOz!eDZ+ridF!(8QB~LSPdcGYyIYG>VuP2DTVl#c5B=W^~HE?b6ANqMFhrbs**)$>b
zS^T?3_AclrclvQ0VTaxew?rz74B3DQ_v0mtJU8BDi3>L<?}KRLp`==0L5T(RT;0@-
z6^w|PSpUNmnec$@#3d#_ZZ0;j6S?ZkZPMN+EBL2@5<*grkYRWVzJ08vP_?_Pd_i>H
z-8uKzo0BJTb94|ntiD24O9T_HDwj1p=tfUEU*ebgl}aIEUMT%)roJO0XuIEBN?+wq
z8~e7>$Ky)op@xmPo*s$L_dambwt*sd_5{uS4pExpT$-S!iH(D<V!c@%o%*o{WA1IC
z^!Q*Zyk|vhUp2ZK?GZjGNp|^59d-|PN5<8Bo}83R+a+acJ$swZ)sKMLZ#kNCdl=1~
zcwbi9<qYuyD##2Eq936tw7V*t>{>d)xrZ}_>2xq}czl}nUF=T<fxXD(X|)s**bj2|
zj=`X7AiuVwm8zYsfV0IsQYnttb)SKLy(?(Fir8I5Zb#<23;dc-6!WbyW6@&2G}B`e
z9dGDLie9&5|L5qg4!q1(Pm?pZ%$^0)^CeIlUcpvHSs=r31kGC(!`Y}XjH?gCi}jsw
zoQYX^ba%>r^$d}BYeZ&kDe7*klm6KSwE1v76{M<|M-1qkZ!!Drq@|I?SW+^Flj22s
z`0^a7J1!xAb`NTOmm{Y>Td<rTAavqobhxINZ~1Z}`p-N@`(2LH%@YMs`|^N&>mQ2M
z#j2<i^VDB*8?oh(9rg7+i<wUU__q@u5H5CZlk}5tZlo@3t**%y$F(rEh7xi)p=Ewl
zI9%7yx<eOIw$s&%={(P>m{v~^-1JLZu<zGrmbc3rR^sg0ch?~NkD>F9%ef1~cuOh?
z*;y5`$x7-u*F#9MNl3hfvdJc-y=ZSLl_ZrUX+Gy}C?ztBGBOHL2q7!)`LF+c%Jckw
z=iK*oeZOTJZ7{z*7nA-ACfEK;*jar`nkeoJBWCX<vwknQ_R2ijQMYK;v1=ZS(;5Qj
zWlylvYCAW%kxYB`%oi+#RQz^N#`hEYWT7Nxldt|!?OA1txHMgyW53|o)KXS=#RnDr
zQ)$Q3C$JNIg6g&xXgqQm-PJx~Y;Uo1I8)DZL>?+_K(Xw$;S*M(>q4e?{zx|01(MRM
znOK!`181$XY2@24Xg_`cjTetV>AVAN&hIJ<6<2}dj+}4qTg!_^tfeKkA-q^Km+Bq~
ze(4xf+GQ9;FB%VtnOZmY*z5vMMg76Iw4vgjlZ)7WnYempC)xZP&m?`Xuz^qGdD4J3
z9=dlkjoRixt6kKDm!zC&R>adC!vOp$(8hhyIel}x6-JxVSi*d7^s!7}$*#ZP)z-vZ
zry4_f$YaVdRHV;3?y_zrq1?;r3hhbLW*Hs7_@?7aY2p;Q!mYaf;n?Ra239Ic+w2X{
zRB_EXN5=sB%3FA2f+bwO&c^IB9+Xen*l)g$^i-}RqPPX`EUv(*Xe(7HHL%zFLh0^k
zO@7_#4vli0i36|F;To=w>hiPXe)UR0XTzT4Tya#kqeYq2nq>5$?102}c01F(9ZIQF
z8sYnW7QIPyX3NzsOD5Y~qZaF0c512&;VHwBt2~V#`#FvL!qcIs7lcDkli0gYp}28F
zWJx{`$A}ZlFvj^NJeCI1S+SqK_xm}m+ii)}+pAcg#TU7;b}=S=oQyRS8Yt*dC>@xr
zfasUqM9y05SG9#xCnHe&90tEj(cIkN8db$5AX*EUtX;@^g;nwawiUR(iJ)zk$8K*g
zgY05|ifUGqMjroXY_PNhfnpA(7cz&^r(MLCwF3kfQ83b0UqHv9M6ycz&Mv18qw}t(
zndE_|>_1gkoGO_DRkaRY6|)Q%+)c=1j<~B;t|FfYBVd!e8iu}u*%Hl6$vj<OrfENm
z9WhN{PcLhdk(w(T?R68$@?V7~;W}dL7Z9;ur2g?X^Jr2O=j1@#e|?_FW<9?g(!r{Y
zuA+Z>H`)`X#H;(Y%Hp{y?m85MqTW%LgdMPoCuk*3Cl%RuiP5%s7<4(E63a8O`Ilg4
zb)AE~K@ms{*@uqD=3Md7Q#L(f8R+F%-1(LUlbfQ)JU)_Di<kecm4#GkAiBmIM)T$O
zi=pi-GoJBVMLII_Cr?bR!rN!cQc0cYJH3f!Eb<!VE8T=+xrQ_`Yz&4@ucM;B$Ehyz
zAY#_|P+K(k6?;om-xUlTnF|(qUdH%0-{|Ir`3x)0$R1V{^2N?ETz$h_ded7$GUfR^
z8g<AA+hW_fr^N-y3pe3DR`TXrT@~4%=ZkUfq#s+dWg9g*t4j?vqfuqDohBH1p^r^*
z!Hid5$ZLNmjJ5n|<aw@ujn}-w&rhl(711A=WPXNbYg%JxDrZk?hM_mJAywN7vdT0@
z&j})TTGWf(b2~{-<A>wIu6P(1>f!OvB-$Q6jN)|Dv3&Pkns_S&R@;MUR?t$!ozF8m
zeM5<k$H!yelSUTv&XckO#d+sWE%sfD#e0b<|7lpuADcu=g13LhD$U+Btl=#*XSri@
zcstuCuSi8sxAX0@a%ktHg>-0xFZq`JlHBUj0~UYN@qC98d`p7)<D0)Ek3V*S+&|s?
za5rU|b=(^Pqa|dr*On)?rt$oP8MH^@OjA5}lk{gP9EY3$`6tl&Ys5?Q73qcfL{@S6
z1ik$i#?>C_Qh1ls>`;w4mfy?Ai~4y`-?0T1b^%D_bNTawepuB16{a02!?_?g1gf7S
zb0-5De!d>_=52;aum|4UAHqIf4a6xigP!o8ujpByW`nL5bL9_f;QQ!!f%U=nZ0Ed#
zC@BoU$gD$r-<%@)aq<EhP6e{=p<CcQp^bU_Nf2eX5O(^BJiA{YMoub2T*y>t46I~c
zma!NlSHPC29)rpFk<g4gKqpl1@wHn!Q=6}g>`#A3*_5YAGRs55$?w)&_|KV)4v#FR
zSN$IpW3I9g<sF!CgD9lPm5#4<r+tS+S7P~iEK&I+*<Lr5tzLQw6zb3BY$#)kP8YFV
z{tqx+bt3KFw_9*;Z_t;4&se{kR|*ox-bP&IT0Yk3AJ<~nVKujmjb6M~_$XYF`jp^p
z>jgbqEhrvWEVOz(h(5mE3-1$Gp}sK@ewr$XvmDP|BbPJB9_z^Fdnf74g|)_c{@*3z
zu5D%R9$y)ky9e#Ab1^#q8xB}{N=EmVLxpn=yI5+DQC;06tIc}K&{r8hXBJ{y_B|SY
z<OZG{&qZDJGfI3XCoRf;Z|pqjCOdV$1G3yOtU9y?mUH^a#(Zo*iM%~O+1yFeP}0OJ
z_O7OVM$_nkNf>qC?T$Dvd8ze&@b7DbsIT=zr1#7ao}J$muwGd@=hAR2Q~U>K-ENrH
z(1h%VZ}396OBTs2Ft5pybbrf9gPU^so_&FmFp;kc-8GadzI8*OTUR~_%~V;VSh!<Y
zC~mFF$B@mx*@5_Pu$Wznv1|kO&h^2K=P_jaqboJN4#nu<uX)oxXEs~X!F>Ou@YUxy
z+dqe4=Hp%DY3@z`6tBqc?M!9gepiUBdxfOr`Zo-WlZQvM0$g=e$!V*Yo5qZkb*;1I
zEA#(jxmw{c4H7J~1?6Jq?~gVuQ?7lmf~Ul-Aj7lwu}5<$CM@YsuU@@I+dXw@u<(u@
zcDjz%!3xHm)PfK<yjARto(t|*5(_oV!Gddf*gWM8+ml@=GdVhuN2qLt+M^`4>RkYN
z7EXs!wF|u8drP)`N~2fQiocPaVdc=9<#t;x`I`M7DhsBuX$51j@@SxV*4ZK}cn`nf
zpMh~>O_`H(Z{aftXJ%TPCHF6dOX@<l;lq9fjD6J!r&60pjd!7Wa?8*v@|(8~KZ92B
zOuBlehtyYJ0VQS4G8+vy+9J++dZlTyz>yQ^wpJ@!zx6PSX=+5{SyxJ1b^)sD^Z8T9
zF*s2jk7-KxnC#6y%q%#I8^@}kX?}*UvT$d8^GfMb|0e!xr7^ZYyN;>*Z@@uvj;VE8
zfwxvOkgPif8aZ=_4Yk0aAs?a8l8@M>u_DhED!fJxm|*KmI$52ht5<e~`df9$uAD|n
zey)Zpzf<T~pTm;I+?JirmcvoMHPq{=9jT0TWM)l!pxNaKqN;+RAiscIQWD_m_X_=1
zE~5JbGGVCDho;t!!pCr5j8nP5hi>f&&1e_ay}ASA-dz$dBpFMca1|3%c2f649mWgv
ziuwA4R6MRZK?dJ`Q0#pNe5|xWOZy?&rJ?hYk@#M=casPEG9nXe|B5qW#(qA#zxXVg
z$8p_<pvs;Zw6w=UJnUY?oR35mjPU-A+apz_OKu4lR8lW|+Vq6zxF4<G`Gmb}@WdjU
z8Qg6AX=eRvF3bdh?7-AP=zOsT3Km+Z%{nG29IJ*y!v-;@F{9zSNm;62<B300YnXgz
zC78yKLapInQn|enJxvm6$Q2`zW2lt5+fL!8y90TxjvKpqK99Ob<QNYux+BX-2*tj|
zc6g>?fp4jgaq-bSGR%%=w?-;rysrla-%ep!ZDZjyJQk8hXDsv6$Elq!>EE+C)GG7B
z_!<elez_SsvV7is<Yc;Cyp-;iJQjCLHR<~7cd~1*TrvB|q=KZ?+BkAMg9<(X_4|bf
zd)E)v!*LTlt1cjZS0K~we+;DuCPGfr6E?b`m=@HZyWCQcc0Ms3(^Lr-eeZIIe{a~j
zL5oSNsu3sWPh^g(6$@i>R+ER$J^o}xIfczgK#%Qm67?H0)?N8^ev-8_AJn^_$WT6H
zc0H_U=hkqzj>&`jc@?SqA|FZY>|V5B#R3GMD>c?HDxiQ1*(~6F1Yi694!6rbiW^^6
zAzW!PzDy1vNsklsY`78b82Um*RYIAWC*c0{EPA^{;7`>8vKv##UA;^wb-E%Pv|`}+
zC>shz`GSK#6?1##le()nE%-eg(UL4`+jyH-ZU|;tHlJzi`iofEa7&`U{2udM7sg+T
zY;vu!4&|rI7oHk*l}`=UlC^c+O&O9CIAIu0E7zZveEX+D_f?+K;tgBjIV*~psJ_Sc
zerYseNiH9+Va6A2ih@H@Cq6>2p52jnBWvgi-g99L=B?;N1FV{9sI4>JrPj07pmO*{
zrqVvI@%Y2Uvu}G7*#&JEI|gr@&F~=CinZ+Lmf1AH;3huWDe?M7Kd$$Q(ZQXT#9e$5
z>Qs&IF=6=I9*V0|GFk6s&J_Ini)^-hC13BGk7o;plG|-YGM#pIXK5i+3iDa-H8c1v
zzrW1?iZ->$4TpAY5KKkC!K-EmT5|uB_1-&^m5=zx{q<bv>lJ@=-xo%1$80G}%%~cV
zKa=hC3dLM27d+@=fUZf(Trd9^t?wbvEf0zfRsRNA+=I2O^iVwRmThA%M!cb^UmQ@<
zZ6hURex|$^wRB5#;vY`$B^X?9*bCe4eB&i~>7H?wxKkK}eLL>L#Vi^>gPhrF^YiS@
zVhzgb_{!FnuVt3blVIocg&orl;cvFOQ;~_H(HQMMq?@mUzE0}MNM1k#)$Sm8&|va7
zxe|v1PEs7(BAX$yQSVFg*@m1~Tuvr30fNVAp;gapIv(P>_F}j%?GB5A1X?|@nEi=U
zknXV!KzV{Y()$dg-Cu)wMe{GQ(@V$4XDV<QTZDz{%4pb}W;_|zjv2GUCAZgn#_XiW
zvOh<cQToagqUW%aYClZElZB(`N<kpmdB_*`*m05s;E34=&7%Rs!s(-6_jpk$i#m84
zL*xDVM$uFD7|Q6%voNyHOhTE$0Hi&M!uFpVG4h6+<ies18kibO8X6*FQhAwwt~`LJ
zjW=OQ@6qF_y!6Fu3uJW*r<;GvxRR9$-;f)Q9m6JY@s_9ikHkIk{Rb$|UME?8a3m~5
zS9+lR58h8Uh+UAcz=)rUD63eB_pkO*rC`-*=T4=nu1d5m*Mp+Yx6s<+SXxy+j15V;
zC%ff56Wdd-kjuZfG~G>AdSKxtKKfq^J<b;$^LGk#e0&z9O+#_PKZn}P3^7o1407U@
zV%4Wn)GNgg<M$=tXjK-^(7S;VIkyqeqJY|#5p0b(Yev}bKxz6yY{q%oZ>9&u@muLi
zM*^Bu&9Jbpomngk<{MWw2+o2ve!lXfFD-YtmTw^}&zIxg_Y{m$Sj*cxmSg?+58QXI
zaJXBG{ONZY-YQ;V6$38txSliEa(`Ezu_+Ya#b@4N&t(2RdLy*s6^xI&DKbU9&&>Z?
zGS8h`%yo<G5!^inp{B&w2Hd3PJ1(^1#0CWH*vRMi@SrP3ddR<h1U8O05vuFJZY}O6
zJ@RA>9xwaKe-2m<j{+}A^-<^i)o-@Z+ttbZRL>}qvyY!vs1;2K2QFjpi-q(wHWYEn
zvAlV}658^r6Ss;kqsa6asP9=z!CQW^r^-QW(|}X7V`wQ+_!3&B6TwE08%L&pr{dK8
zp5(NDAC(NdOC_ByL%H8GzQ`w@OjXoaOz${ezs>_~Wm&jz@)55qt;9sGftR~nS(<_c
z&3`qQCEU6oo3>eA>RcLy?!SU5dT}>g>AV?%{gtKVhqqzK=mha@vqoy?r}Wgk0Ec!+
z=<aonS$$o}z^#HjKYc;}pB=1iTW9I^j3H=weVCc5DHJ{kai=HGO^88NT72*mzHGTb
zQ!big&DDJV+&7*z#qEXj_yXh~knypxIs9giU?ixW!j*yMka&a?c#j(e`)LWRf3X=e
zvGL))zI|DoOqY5M6fAwKZnVHLfuE{Zq0EN97?u7Am-f$t{k}LJ@unXZD87_A7`o8J
z?8&^nt_}Gy0lanTMq01>mv<DuV9UgFBHwQ}&7IJNwr`52Y1NVB^eLVO`~Kk#HCO5P
zk=^L^y`IhLa|)I@r|7!+Jkc4Bkjy`P4?n|N&@^r;^_Urk*JoU5dSE-5pGbj$tV^MW
z$u-Qom_(~-KB8_$;l_fw_!RSspMBX`dUsSJu6ehxkXe9t*Di%$8mrjik899J-h`Zb
z<kN^O6{+LO&eHJBxA|;?Q1<@s3$F2b476JpgZZiQHy=YJ=0=e?S#n<NF_LAiZ(|Uy
zzeq459qC_ro#bKgW70NKD!8QQBIXuzXpYWqGC46BdrG#`nSbSMc`rlupHVkTu`j`|
z@LD84y2>S2rV77nC!D$FLzZ3l;Yytp9_E+Xl;U!#mqbZEPHE&zJUVj`?vo7P^Fhqo
zOW3ZrWfZ7U%xGsZTmIGu;huld@8(aob<clTAoz~1Js)HF`Vgl4-#+X(eG6yIYM9pV
zU@=!uL*~`awE1BLukG?%a<!_1jZO4n+xN!O^t%Zx*sY9J&x~UA%J)%gZ^x~UT$7!h
zzmJyO6il_N1Cep;CXAhal6q_oo<$55o@!;xEM6k!c)2K8bDS3q&!RoWVlUo(5T-8g
zj*#ikxt?D!KXf~e9F4b#9AS6XcU3Y|vz^6uF9^j1O<xj905~+)upxC?^eazfj?LDx
zM+e_BGfy|8|Ga|`>0$#%nHze=`*7E|*G#MMKkn6(LNfOOOmkWcmIc~k-QN9px_&ZM
zPLAQ0303@?O&)ILRbi#6=q}BW@(wY_U1P;D@M!=K3$`Hl6^jwr<uuy$&7j#a78%ka
z8mcw{1&531iJ2pPU-*cB`8b%ZcfG>iiamhU>YnJZsgTTHw1|4Vxh!}yLrE#yACpuh
z{9$ZQ7QH4O4j%(>K0caCXU^rt4;PWc&s>o^nn^_})$m&B#jg&Dr7FG}4fPwLlPB0i
zy@%q?><lbge}E+FW+=ZMi=!XMQ~Y?r%YSwb|5mM`-`h(taq$^eTpxiw5p$sT=(D8r
zmQo3RJK;*25v7b*mkpnj2geubG*@15wG%I~=;tqK#?qh2JClLwp_4Ip;$_Hv4<+k~
z2~Zooi`%Mp!?HmcX!Tc@J}md7wtfbfvi=7fs%3%=&P@nvXo1ORb&4#Efc{){1lA6O
zcH=8nn)Zj8xBNn{PjxuZ(Tdm_&U$YIJ>0FxZE|LklV2BF&9WqVVTM%yNe`uq;#lRB
z(==}|7rm47+`;PxIhLFw_4_hDWnMjB?b`!$?=PXMx<p=|sK}kA$#gySH?mIXV)(kz
zJb2I*Ozlu-6AoAM!t79P>Xa-vYz27tLxzfl$#6TE&G%@#Q+E3sQum)lfqgvrSLZC&
zwY?|Frk}_0v~aS0<4%F&(~<w_EHjAxff+61c!^h>*!#xthkewsJxT-6Pa@Z?7b)&-
zx$*cHd)dUtCs<pPD~0C&WMy9SaLPcLRx~=%flklRrDO-`tWf1H(kr-a|Cr(@&jwcb
z;^Cu6!69EOSn}rld)r7jGV%St@f9P+S&{aoEs~XP?#%S_XYSfDpJ(2Egt|5UY*LmF
zpOak+r$ZMo*We@K?>(c=;ah3JV?SmOFO-=S(EbJ;_<p_%HP;r2%z3ix>7;u&>Z5|o
ze(Jb8SKH|OjWjY^T~knKbe#^p@Wngfs886@hQab781wlzem>8n%Fb(O!~-un_d9?R
z_4@J!se{ndIE!xf2;l3#F6N)3LMZl-CN-J2N_5kg(X6$qROH(S+dj;Iq4_72j*3Cu
z-#X}|4`#Cxx3b3(7ZINo#72KRk1ww`@Xbk`U^PGobur?e{HF{@M&(h{F=ykAbuLV?
zW-fgYIq0DB6uh1P825G^mz6FO9y0Uu*kRK~6}??x_#_b8RxkLa!i|!RqVM_3!vj~B
zSMqbS{;-TcC4$ZSm^OF?vHNq2$SyZr;(mG^p4B$NQ2DRK>Ge9kb!!E_>7J)MmUaC0
z&Pd*9ae-N#zQGq*yd#^V19{$XQ;L341cP~b(AnWiV|+zlXr~g^iLT13zGl38=oPlH
zQ?8`qlnOFJPQX9a2^(t%%H;BzWi##8pv^OkN<QV%p_#vpPb?6O<v~}m-}ET0vt7Z5
zHs3&E$x9YFEQ0kc2!Q6b3MBRZi$k@-;di~0)P7M4Q#qwZ(zM$m)8|1C<3Gz>lOvd~
zm}yQ8)u$roz6D*MSD`L7i1ZDn;Ye#VMtk4oTP%dJdr>`~QKL?N@i$;T#SiS13r?Cy
zanW%CSw4#qOn~8}sA-IbM16d4=t8acTlv?r!x&twhGNr79`N%rG=?nZ>i_vM4TS(G
zX};nXorch>LA}Vg$`9FIQsmrT!^T_tp>5Mg$==cpG-OqOo?Vf~hw?C_ktr;W6tZP~
zRbf8-HPg=AFZsCHAGOLgY~K>m8MK+rmRS!YUH?U7ozfz@Oov!oWiRQo<?B&I5g7Kh
zQ{jR0B1`7rE9MFjc+gpa8P~SM;ZZ7k_qUV|MaHs|{koCywlPds;ttb4;_T3Ot*o@_
zhs-tC0UzwLsCmdl+V)WF&TUiBS@Qy>JzvioHGFvO?vZ%fw;W?Ll4x*K9pqmJVULL~
z*X<R`{&|*BXocVzsP{+Ox=ilEHlz4<4Nh*bquqA?EL2@t<b0-L>D6X7G5ZE@KBg%=
z57{(GjK($eMls)ZmBMddG_mykaA-u1<_G5lpxVlsR(21hYX^d`I>`m^l}!ZO?G4IW
zu3=beS84q2uGA8hLW9?2Q@i>}u_Iz+-meuir%piQ*y(h@;X7Q9DkIuu8dUZOKkxlm
z>gO^Sy=vd1a;yv6VLF>)zmI@nj5^eRHj!T1Q8I6^Muo`lU!aqaQ}e(U!DPspGL~dP
zgpWgFur;%cq6P(0y-5H++4B?5tvk(+9d(7tDKVeLALFt<+xXR|wR}=hn#gIlL)q~Q
zd`cowzh(u-mR_YU+jM!#*lHd+AsUYapJdVKPEt3saul{q;%i3eOMaP*q&|C><K%ZY
z_PcX9>0}R~O#gfI)8qo*`s4zuXb{}W?%Qy9p$9L~4aeYDM)cC|q$F|MBr=tKK}*k`
zP~d?$u-}1>`JLy*?;eT$rZ$|*7r}AkXjs=5$jmcG0)jcGw4*y73|t}JQ`_M1&{#Nv
zYAD0Z3wFnPW8^R!MzOCYfyY`4h8jLV@k|SJ^o_=#6VdpUZpU6|m>6$S;Bd~E$%o{x
zVoRUTA+?3YY_g;av-dhl9iL8O+n28Nx6PM%TuDH}!sVj3z@V^tEt$C0;+CUW^+ekU
ze}E0%TC}tDB007-wio}iki)w`xXVPgG%Zqv-VVGDC%b!;SlypOUggoKg3H1edK7<J
zvt*b1iSM-TJ_@h8M``6p5gFgam1}#`V>x3QviKk!JDz}6@prG><-rG*?ZcZ-!7QhD
z4{kMTD_ItAM(<yDFsZp26O;_d^T~C%*n6VboWovpjo)UUW@MyT=w#=Q1r8^1Yf29~
z(zOzucHWSb-2O;D|4rp@ma0l`?rva)W|qQR<tfQ;E@Z0;^H5wp3KnM@WE1<e2oCHw
zdQ%>Oge_N)uN{vyyNvLlJpjj}J?W_WF}PF|lY%(M8Et>h?eF^1u1R{-tmuw|m)sFp
z^^)#u`Jq!yEG;a*j2ko7LoX*&vZ}oti5gzauVsYbaz8@GW3|GA*8*f0m+L@z-X~^K
zd6V4__F?<tImV_v6fD^yc8gB2Wlq+J8^GZ-?F6kuK*9JVck=0)!>@eyLgu#*$R09-
z4Lkmh>%1|-<|S*{sx*<263+C-qgQyG(Hq=SPeI8gBU<2jn5l*@WZUm{HO>uFWzQE0
zclk|{XsZmSUrJkX&+s<X(?oX9sF4M>d|`ZY3~iAuqmTJaa(~Ae_E|rjUPdZPXGe6f
z?0=K!@D9%G7F5D;#|j$$GMaXcP@wyA=CWyWq0G$gHJ|wQ0e+nS#oID2kdkl%IhOB&
z%vn6IZym;$c?(f~$BpzG!)Vov{`i!!l%9N>CfToUhcjVTtZb(OVs-1`lk=6oR@H)5
z$v4@S&;;5e?@ngdlc;upKZ^{HN5*_bYWUHW8mHf8nZ*Hcksp9BK_Z_%CI%*}CeQ}M
z8^~GLLW6ZT^7TuPa)ZBLCC(}iIQdq%`Px2-T>WI>sfuN@w$H=3p_w$S%Rq|UvIS9l
zZ7E~BHxuY-q=}w*kct_#S?uO!I-P`9t&LAQdIDXI?2*@5iHdxBLVu<VS<%YO{@qmS
zxabDin{_NH&s6qr<V94e<P#o!XA1Lt$iG-c+6NCY>E3WCe)-F*ekGu<qc!3UG9lS2
zc$fLNXw~a;$QuN@t6Ik+w6040KPyR>{CrfPe?f~X4?p6HYCWX|y`$mgkw>i)>tv%x
zSW@?nud=J<9c<EyD|o-cmEX5tMQJmIYfR3NO)_Ti)^4QkZ|1=C&<l1$Wji&59D{E9
z4#{B!;T9UJLj&9VdENTK{DA9PI-=dq);``r1*Pria=ZyDDW=8)iXv$6<mY6lC4rS0
zK`AkUEx2w)W(!yIRh_2uzH6GvySuZP(N!b7?;-q=>jG=pUm8Cli$C?c$iCdY&YnL^
zz$S^vKNWe9igpxL_^YzUO?Bw1W>3ZqX<+9+Qk~ezr%oJ72ExVYzQu(PJbshB%Ld?K
znuJn)%<08ucjMiMmr%yy3zU*&h78Mb#MUTLVp;+>4;W8@8QZW|--}G1cgX(5UW8(g
zpKQ9m7Z2_3fMwyIkgUbnd&@24lckD2^<KD4lbNPbAe{O{;jy~-T+|Eh?Wdz;R`ZD*
z&WlKmy#+m;rcKvx^`Y$#Q|ZIe7`)A}#VsvGv7fskF^hYGtsU-=jowEm%QD%iF6o#)
z@-WRh9mM;ZIpfv1aPr#t1Ql}yZ(ca6?xwfmYT*WozVnoQ=zIyqmp|a|uy#qB;%@w;
zK+=1#8+sMp=~u-#n#7(UyH9`0TCqqrV`Dq}zGMeq7<HPdD2|{TO<(wbUoWF(%VD(l
z%oADjdkEWn5{uVuf#$pdDCj+veRv?4`-^)>(~qyfmzDaGc{g2I?7v*$)&0vRURa0!
z3hEG9_#Z5GW#I6m0Q5L$h|Iq;$X6!<19SUQT?xm#4u6y<--4`34n28i#Or#>414d!
z?~ya%$$H|Lc_`d|wbSu_2}tTYnXfhg8Yezr9Yrblp;aULMsED*taI@1+lf2ul$R#%
zX^>UipGM6aqFCC*yKv|c#3MVernh3QAAb9W_<yiQ-ICq-E64e_Z+7h1lpWOatex(h
z?j&0@OPvOFi$>%53A8iZ53`Ciu%T-h>}>{Ne5`PsKWfJ>|4YD=4)%EKFlw9I0u`3d
zo#&+I8(E#EBQ=U-q%xl8Cn`us9ZnRUg550cyCcnZSD{rr)z~C#66&)G@JAt%N3BSt
z^JXh>Osx!CKNX^)V-rQ5+>LEcBhg#jgMYNV<{v*c!8rT`9yP}>r`|)^?FeT&RJ(}F
zjen9qCc&Rdzx?68DHWt^9YxhkKd_pfoxp0fM3?>!h7PEt>s>gO4DUsT>+~gO{0EDE
zgYbpf7-C6FjBKoWA-j6$w`_XmVQxA&f@yTmVf}l_N&gJ{A~JWWXpMM_SIaiiUK34Z
z^=QS9+<~O>`V}Al_Zax~25MWpv|!zx2<WNxWSx0edby#DKkjjnTW|ECF&_%JS??8m
zc4$770}9w5ziX_eu3B>Lt2k%$JVgFKH85Q4Gh(*;l7o#0xraoM`O9M@2dBL9+vb>9
z*#oK}DpXpxhGzDQpu>CWn1z^0ty>lhbvt*s>I*+cKoW(&7kJSxWy~`*UlLvF%q|#&
z(v&|_cu=qj4+{;UiPg6JbaSkXpHIM?BdvH_mVu-`f=ilggb7umUo-Euc&^8@mBW27
z|L#ee`qe<>?!53dP)_=OyWk$SJjT!zHz+?_gM_!&(LL%SY~EH&zDM`wc1n4|$#jh?
zk6TR{8NYb`oq>|DULj-~VZ_fo=_D<)&7hV57Yr-N#PArwSB+Rglag0KGCB}dtdIuB
zyYZ1<TUcg$HMhQ{%AL*K@bJue1a23-hNpu`G3*7I$z>o}I*RP`Jzz1|8=>=E*y&Mm
zSTo%ar>iYVf44Ipj!<RQd&8hP;0~+z+`=`F&u57V#eB1TI1ZCH-ktg<^0V<|9d`t5
zgy5aX#bBhX3#qpWC&ZT}xY>mv^*|n8bW}=Szjz{8Cb5!UZ$gl4?ak$qrz1W3DSNP@
z1ou2m@I!cDOpiTbn|hp)oGD0$q5WH0p>r9A1K+ai0R=p~_cb!<ISea=$JXgrG~4Da
zkJUre`DgC~WZ7p8*L%4VrJ`$?qBe<~yUG`iO*#oJO)tDN??(}{s>r?3QDpV@OKb%f
zW}nj)m}vOJx8(zw{T<01#=9}awsWuu>I?HT+4RIf4X+l+k%7i8dR(r7GdiEpUO!W&
zy(W&852|8IcZA}Dzu=(<3h&V`;R5U}`io-^h)jPdMf}NtP2V+eEee+HF`Wk2iA&M1
zRY7?2+<4E_GQ2)0_y^@}Y-XZ48<A8b*sd<nzAf0D3mWiG&5zC2+JVzE)^NMc&zKc6
zAW6kUo@;)RN-LEL(+-s3-<JTM_FyD4IK;6>vIv)U%c$hea5(S1Mn5)Pf`ZmWu_xGw
zeM>4?vY3sY8KDltF3*h{ayGEl&ZoHEEPu9qaR%zXPUUG&G`JXBqN$4)mK47wx%lvW
zr<{ja=ptkP{Z$t8e1F-KP7KO9%UM9~A?Ce3gZs85NWxYHQzA_j4!kV>Qt2+HY_CUV
z+ce=I&M<!XUIkM`Ki_%z8r0f7g5$?dbTP7sMzp5L9<HvZ*j1@0cv4GyB{`@H&Y?h~
z>n!z^BljP8N%Hq_H)(cqGj6*HcS7hDx~kKe2k;cSFuE%O2Fcj5x520p-E8G=0qCmq
z2X?5(^~pK3)!s<-j$HBVvnTZCe`m=R5!_J!D^=^|%Cv?}k=X`Lq~4E|G5AdlJN>nh
zc6u7hS~?1O{-KFvG+%ILt;4yAcPi_vmk0GD$!yuC1PuA^PWo4~S>h+b857y9%`whY
z<#QB`X)oDxrLo*`Rsw%K-j`H;D&UwLL=DL?NM7bjdK(K_tl4z*j<sMVE-jc)^G7xz
zI*g|+7S6uis)7j<gC7oFWZNYU`!0NvP0`fD>~JI2H0B!1?N>v)0zz4Er%Pm<YY6os
z!K!<qisT;l2sfX@6<6<MeOgNKX~6*;kLe2QrDrf8_AG2p_m|CF*(6iT{=>%n$)lZr
zO-RF~g=d<t#AL-vblzNq&!*LMy;a=PHlAk(hHFTpGi~^$)QK$jxCJUNONp)vHsypK
zNG+N|TfDpBrECLs&b>j0=SP!HH$~}zz;nzdcMXzcxqPkjJ-p-_d2ZHG(K#<8L&HX_
zPkP5+O;e!6jk~BU&K}p7ouQ5aEiB=NH{aq{&%d4<gD`&^%Iy_Fa%5uY{oYd6{$M?>
zEl@(#x*&>KbA{H8w1@sWKWb(kxG#F=heOWtD3PI&3|FCIOL^XVQ#~GRa0SggfkWAq
zGPjealri@=?=kfUQdh?D=iNWS|AYbMOpAs<9`p9)ZxOs+JlBoiBK>f%?9>`}+#Tyg
zz4a^DpR7CBTv$pC8P9p3?O$GfsgU#^>aZbW9w2+66|FZqgn4ll_@4I!mf!D_HYXO?
z_E2&_%+<%8sUnBp<0M%#d!pfp4gGgabl`6+V`alE$mh~``Vw^ynbuPf7nmTi+&_T*
z>0XAEp%?M0Dn|4)<Jh_8!!XD_h>5u=&`Q0<j$RZ_%TFRJU#%?m>rA$DhY87Lu9StG
z(#9(re^#<lNt#he5cY5My5LGj<sBq#C(qzs$1djc<1bvy?^96id&!eZ;rt#YbTEhg
zQ9MQqRnekPXeZbz&U<iJ%<4^t{)fL)A2NsbEPhjDI?9F}!j+2}(&d%%(y|9CQhHKF
z8L#(CHrm+B1}Lagc#qCxCeFILpPwTC@daEi-Adi|rcrRqS6JT2!`-N6roF{a_`}>;
zpP98{j&cr8+N#nh{r$qlT*G&t)sU<^LCk$k3bHDW@Z8ZdIIOVb#lNrc9n0S{yYysA
z*2!g=4LNM*e`Wao`WJ19T*@_q-%*EJ6)fh|^C(+|LbX&oHlwkF4Vbfu9MZyhUj-TV
z@HFo2#>w3Sc$4;sW{uxM##VEnEOA5riL-Qk@LfuIxQ0|5EGc+WI(lzal4i>$(xhTJ
z>95m}7`?s(>vVGpeK49d>T+>#e<EGJ7DhuqSz&EeG;{ZjWRI6LW7A|$T;FMly+y8K
z=R1dg4e%G<ola7@SAm#f?*+@+Nc!zyO%Fb6gQ~MB(a?+a7^cNv1UI7puxH4x{s_B0
zZg`WQ15)gQ{plW<C^{Z#10zJ<V+|$6KfxqdDRfq~k#B=5TXeNk;k)QS=sudwv@fQk
z#rPVlpqa>=S4c_PUocfO6M1&m@M_pFvVO7~_p?`X+nO#US^XNNmuJEG<|T^K9!h})
zX?S*XAD%rv!z*=W@rcO<D5>~M_RXiL)6bD~-e3l8qa>X7bmYBHMo?<rAX0u*hRI(A
z|15VXA~OQn(=or<qKPKdr0<F*n;>TUZUXs)+u+K}OuDo7B933zLe2m`X!g!RP|_MU
zRy9TvR6H50Joi$^^rhH&e>8PA?uQ{WV(692RXCi!ExGR-kHI$*_~@ATJmW_!uULE^
z8}saOO7tO<4|rm{{&`&f+Y5dM3Ak5ZM<(ymc*C$y2>Mrs9<CKQ(d{$B3vST5gTmo4
zsu>DcK^yz6papx?=*4qF@vI=q3%`pEMqQw#d>&)Z^?}G;vUa@~!5Else0ka`Qa-7S
zmh%lP?c-f&{?Zoi2_wjjw4dJlnIa2(=`Xn5&SX>f0G*cB<3Hi&+0bnm@>jWH=a{=(
zv+yhOR&#ohepc4&-3~tDYbX*=+(h5wt6`QYa`<QD3u~{Y@#8}#qQg5{GHHh=)sGgr
zFZ))>i=GWMUTy?hUn)`lY89z(Ndwf*93#t=zPRV5&4)}I09U8;v@tG)*2mOK#@Q!h
z$ibOB<jY0L=O1ZwU%Wd<>zu=rk_otmLOi!q!`rXZ*@4v2OgZ%+t=rz8iVx?JvR@3^
zT^CE#_GWXr(Wd;hw2}O-reTfy6#o0PCd0rIer~3e8w`+_+SvxOGuH30vu73=e=?@{
zkBem+FRbKR+2Hp+=2C0J8h&iWX__T6`Qb6oaMW!Kxo6y><wa)Xm~cltj~|oj%K$37
zQX{Er48<zBOd7fA9aanqX2nNO5EqQt0UsWa^|YrFIgzh)PaZ<S&oX3|8-(}$vlM{^
z8z@*jQ(ty?Vd&VI<k+pVG&8aguiol1&xmW-tDAs5mqS?lzccLM(g3P3d5e#`h9mEI
z3YN|@LGyqJZf%#r+O$4%Ey3n$ADDoTbsv~^KtG7vD9za&z*H6#k$2>ImX_&7Nwqzs
z&M`H-r|Wp;`f3VJ_X?IgH+zRW{oZ2hz%P=PFyZ~Kyo)sZ8?<I#E}L~Q2|dHB@$maO
zI()5^9zSwL(!tYQ_?4(vlq0Po3%Yu1BnzDYvCoV}dO`-J<t1TnR)*l{+{F1?K~&;a
zj>*#wz<E;-iTD0LG(P17T{`@bbQ&Vi`BpWHn?ILV@48md5-ca3?PS9ohRGLrE);y6
z5ieLn>Tz<<`AUs%bJ6+oIz%Yw;B{Cip3k>H$*gG9AGks0mf;vPBLkX`d?hAUYEs@B
zf{Ggix!L3M{+^fury~<FINlXw2i0KtsL6cnuKCCj?63q4d*Xv5as6K~WO?;`_Wn`0
zJs^vQosDGgK7Pf(v0bI_M=DA)p9aA5;ZuINO0d_`1eYuO0s;?4;fvR4vY6Mx&h881
zp5k}h^;LMs9farCVHw5kZs+!1QIa7Ybxi5yLG=42GU{D3SkdElb}7(>o=R)^mlfil
zFKD7yANEo9d@0GyN--j_7s@?aL=RQCP($}%PEIFoQq{n2uE>?7PL`KG7=DR{8@*$f
zhb7Y1#B0pQxfwcZR-(-C67<IQ!{PnL^!tPxUF#fxaH|A<;@JS~l_+vsdBFk*TEr{o
z+2V+thE(lJDi(Lifv@5n?xB&&T>Gcct1e0G_9^+o^Wy&Y_Uu`SZk{9L&fhEe2wm~H
zYGS7cUxIx9Qg-#*O&q@7pY$`A)B9cvm|@Lsx>+!bUVjVWM<x!$EWeZJHFi8o62dUo
zegv#7p0giI&ymu9YjNuRCQ4`vf_Y#pU$EhhxZfOQ!FReyTV4wutcf?m8zqv?jho?-
z=>(mTo_ug&HBWGQjLeXu^tm_V9Zm=6-ici5KfR45y$_@@54*snvp;Q>eTHYJN}Sr8
z#?!TSVOB^owyvla9!>+<)0{?5E;VfSn%QiWMQ^nAyMx~j8*uAvIYo|~MvFY}W8Fwk
zq+C8G@^8wR{4E|vk=MC-w<?}H{UJr&S4NlGi=xjczGul<EMxs~?D&3;8D5M>-V{e(
z`+XUmkFura2gXQPeFy3h;w(4T5sM68L2Hr%Z|v8?wl<mKSC|GoEw>{zK8r#RyYuDy
z+_;un3in&N00)>g{ds>JFW*JetBgE^FDWCV6((rtmrAqO_mmaw{lLs-H467mF48JS
zlS1EkwxRS2VtS3BDdlHz$~KI&^qgSs+b(>?x{TT%v*NvC=CEir4R2XbHebexc~mV7
zS0vEn(gc=hb`{4htN3*Nbl5(yp+55xFnaWCnq7Pn12#J2(3l@G<KE(2dPF$Ie-@*V
z+hbn+etKSbn%@5`V}WzzjOAnRNrrCijGL=>N*q+P*%?zAovc+v$=pbcbXJpwI0W-&
z@9XHTrHp>AQ<To~-hk&xv3x{IyyT(H09?P`42`V`Y?87C=?VtOFfnUe-npl=C|Is=
zPcLiQKJ*UUtSd10=xJJfRG%Mu`3#9KzB8r9-_ZEyOk+xN`Tf?@EJ-|{-f~-(5Rt|F
z4b17i*EU>mvcsZ)jX0FqnYC=RW@fDuA?ruDv2wl0toy=mtRKw}h#*U=&$xIr1xk}g
zl4X@U>>4g3Ccql`ngw`ScpE)`7tqyqj{269*m>l#%qNN1z_xcZQt=!gqH&AO5*bd?
zsbR`JzOi(lFgPwx;PX$U5%-RV{V7k}-|j}q5gK@;dlohG^I>)IId*D&MN!@aNz9}`
z$g<SouX~$4X`M#p37NF+-+ny$Jc+hzUV^uLG0{nVmcKYaxHm!tD=L`$7u+S+jR&!9
z_Y~Z-S&Ged%WyyMIetA}fP}GX(t}q0$--qSL?D>U?#E%-=_=8~R+28-_5yu|XJC`a
zQ~f-gO7ptzl|{QQ#?_~BG&U@r<u4T3j5|8iv2Y2wg-t{0Opc2)e2{kGfyi%H@dKnU
zd%ryr%9*OHDDe!8HvW(t8MPmkpHx`W=X;nmP>Wn^E>idE2IE|ZwW5D-MDIQrQR}aj
zxO~T*R+L9`pQ+cNeef`=nyMkSx9Ud+)-|Had@w!K)S)@r7r4Gx10pO>VtD&y*79C(
zimV61b;cBIeQPXm2_mZ;^PN3DriG$KYxBSB{3^Jal0wfOUM0)5S>oP*h?-<SSW(Uq
zzIsR`yjzZ-Slk2L7YEX@A#d67z_HZ4?gLxmS}WN#D~~Ch>O}pzWI{>Iy_}Df;ORfn
zOYNUc2J?F%^5ZXhuswp3e6O$uH<5QeDL5OGn~9&A4)5RFNnyh+I6X+G{F;?ir#=|x
z?kY%~omS)f#;xf8T!+Ri{l%{SzCp|P>Ed#|H&hK?3V!@m9v!Vp@`^F2in~P-X}PR1
z@FM+G^^_ci79AdEg(9Wd^wHo1X)C^==h|;s$~VEbyLSVIW6n@@JL8Sl-Kn)~4l{qk
z+4Vjx63?fqST|klt@jqN;nE;6gA{zxWHqKYITUKTmE5;7gf~(azhSRdsM}t_<1H6a
zWJG7#wz^!VZ`?_$a&sldEgi`(_ti%6z7R@VRZiKaw`75R#uhyM*9&*EGx@02RHmvs
zhDLYx<%=$c;MuZY7_GR6FFf9rl7DSs{SJI*wp|sZCwnYL>HK>vJu_4AEi3r}UlrOA
z(Jt}HI05g?vFNkB4#6VJ`@QNOTbdmpm^psr|KAVvovci<%@GpUmN#rgbu6)yL3mVS
z$zL8mhE^Uk{er~__CULwDzYCGWZho?yNDbNvH#6fd-x#Uw3@3}ddp_-SuA5M^T?om
z6KVsjC8KQ@G5gpPq&c*J4YbQZ=T(XLvM2(kZlBoB^a}psp|5Ph;y9$-&Bcn&iKuXU
z$=g&dWcRN6Kz@oRtvVINW{mS-aVs}L!TSN8Y&y)$V+!zMrC{8P%t?o32KLCZaHZD_
zNF>VAPaDrt%4j)h!rTzb*H5O(zl=?*HKARW#pGovI4?UNAksdR4LmuA^r~)(eNGCF
zRL_w#j0wbx6)*XuS69$$;soB$sgcdgynuxKQn<UT;ClCG!T~#s{O^R})Mp*ZaE(o*
zw{|Bps|=#dO?R+${WMzTD7fGa3;Ew+&SbHng}o8Gr5BeAu_}lg{g|2zSxOjfeW1X9
zeCSJ2($(x-cuz6Yj*%5Liy36rLwacFB8#f&NBykM(^}h6u>7%wN**3$pJ$6*%22|f
ztXoitdO*+q?xNr|t5Ew(MP@TDh^)uh7StQ*(2uir65Z0($RBf0;;?Q59Lfza@~jQ+
zu9-l4o5Y;d=&G#m>u^ceN8@Qv;%LH%5vc7sAKEYTCEb_ULHGJy=q8-U-NIMKAFIpZ
zBXTA?R#=iGUS4`zV?J%|WI*=4wvcBJ4&C*|5~swC2v$Eq-`>be731IIgOxT0&WmFs
zd;a6a_l{t$sRqfHOp%38`vK2J!HkMr2Io%FyxC2sXvOGe-c>7(^t+xz^@G9qH?;~?
z{yov#TaL#B1;Y366li}J98^=rR*kt-z{gkfN&_vr^m#IWY1S3Wd(+rC<x4c~>P754
zcoTa<mSg;9j&Z6ZaB*KaViNxH%-5H>+5JGe*mjY&Iv&D}ZPq+Vb3g9vkcV`>5vySz
zG2>?r?VA0BZi;<XLZyn-VNoBNGwTW~2++gpn00V{_Jdi?7fuFwFFgLEMDKqU;o#Rj
zWRkfS6CJnVZQ3EYjXsQ%oBPldiwH!{xxfy{C!?d~9GB@>VOln5cETBy8Hciw|L#*b
z-v^rJihmpPDZP0cCC@KnPw&oQw+1Ron;TuwzH1|ybW6fT{WNUe*AFuk?joq08rRUe
zgO)GFh;VA=XAcIEgH;pnu1MtW?M_phKXaYGciFYk2XUcEhg8z@F(6QQla*48#~zr%
zd=ka3Cv;swg60%{E_w$3_P2qv$kTOqS1QcckeBZJdy||uFGt;}MNq!sLvdB-P_uHa
z@PEFC-0oLwbagsRT{|HpG#ds>C(+nG*YNaRJ9^aZ!sH&pN&9saO!5oSce!A6U1uU&
z_Fgj7OXTHT8#p~%fvsC+@IFbs$zN~-R@CTFi+LU2V@U9v8c9*zu2WQt1%2aoB)3e8
zPow?t>5Mzsw1i+-Cs%41AlR>!O44TM2~gbYgW*w6*(%*uS<pc>-eKOImA&@B&Zx0`
z-pVLU`Lv(LnuL(gwJee%Snv=Ia67SY3-?Q6a%LHDcksZ35r1Sui+>sW>Q~{==yyzU
zrU!<0Sm20+(>$N<g{ED8qe-WghRvOUmP9QI@cN1^qvP4630aibO_2_DPDj`COU14u
z3oZXbX(OGNI8`)a$-wW}zCW6+(hozz+*ICd$yt_;D)ev*NkRW;i>no0I##poVjgd+
ztAW0kqcGBEF>SK?fEjiz%qKgO`qe5(FPJ@JsV-?$IN=_0wc7cgzYIH?omp*S0DNAh
zVxE#MFI^f(>ju|j;6y_<WveC>-$XET?@;IuoJ3i<U)i&5`FQ?j5gK>s;X$|(%oR&0
z<%pO)#^qzZ?{#>nCes}^Z^?(|{rpa}2a8!!&Q3{pvilbuX<cU(Y2b^)l9KLw(Wj=F
z_mLL!uJVuBVTmFh6bTlW{XiD%&ar!hH?vu8&-6cW9vD!`t*_mogvQGl=-((gTh|j7
zw-y?=?dgLon|M5r97F3fwsDQCt69iqPv{PPgiO{Zx+KHdnC@aflM{=@?&^r@m5AhS
zwzSE&oCjOHhu`d}Sf+3ld*j;JmseX@h~Whah}@4cvIDqyBt@KYHqr9L@hs+XCH_kW
zmE<l%$F*|Iv#aEMRUS^9-9poRUa;uurv=WPOk~rRmD2brZxOHfolQOMg4ruhLB(VM
z9L0S@WrZas*4)IksUea{y5pE*)mhqE;E4+n#W;EHFKwUe$`aN8@T!yP^iNuc9u>>!
z>CjFz>s$}sI{Bom$0A~GA3w4636ChbIt_|(uX)8T(NT>(fzj}zv{%in{GxCU*%%=I
zRs`BNE7OFQ@sLhFPTP{h@jW6LrrFzNfAW4IC;Sv2q#MSgcJ>k;W3XI{B8r_V;r&C8
z(y&iTxL*GX(&0^L=rV#wObUi|lna(rK7gsgP&%ZMCFwKkC8OF99N%>fefqk~oYY1l
zY^~VAb&EuMCr$bj_5xE59^iX!%f$D132iP8ktuu)#O)<L@wjLK)K`dS<%El<DDuKb
z7ai2P8PU$>44PYchDtq!pU}{h{60lNZ(A#t=*06r_s`>KNpC34UIzUEv$5%&;0!ud
zvduFh_|9#9)HgH$lb&9LN5DEP8XAdh7b5ZF(<OSn>pQ)g@eI?x$uLagJNs1ZMBOJ8
z^MxUIdFOOj(H-$-OY-CSn2nM2I?tIV>)yk&2l@Qy*I~%)zKzc<b7f`y8|i&>6BZRe
z;-z{HFlsB5g(ctNa@YFOaH&5Q^}NOReY%X0Tf4|+i9LJ=Z{^dpyHj9yS8^RbjatvV
z#D6xsG1+k#H#w4nTK7EM8@U`&rapYZfxUG8Xc>+9aU2an1$gwx2iIB;38#)crsnLU
z$h#M)*0<C+^-l;HG{-Xi5$VD?bDU}dE9k!YQnK(Xpk3v}t{rga3tgH7yUvj*Z0RQM
z?+-B}HH~_2>5VH-3Q@mYgDy@OLr3E|oi#1`pQ7`Q$NGE2xU5vtE|vC1g*5P->yeg3
zX{$s#jV~<?QAk3_h!9e;8;CNVbCaZ!G>H-_m4qZ|seb48cYk=r^L#$%-1l|8-}sT&
z#yT#i(SPFav(;3aX4l@riv3MuR~^9Gl0@G%yqQ^lcjbJ>bqrqbOb3_xQ~#B3>Bi#}
z_9!(Bs_p{$_PHOASDi?=`;3CS#zFpVQz!=I%|&I12lYs95&hP33{z9cJ7_wHtQN#$
zRnIVba7-S$ue#?YHHOlX&_l4Sw&u1;8DNhd3U}>i$wu26GByLo{qu$G94qcrIgWeT
z>dI0}6r|JmP1u{Cg<0`747sxbEB>>fXIe%0(VU8LS3}^@t+Ui#dXuh-eZT$n&QjCP
zf05fmN4#IMF|*$mKJ$Do%^Lawi4HojvFe#O`M?JHlsp#IYv)2JaoN`m5twl?g$67j
zy0bZqr*-z^|1DA%p2@B-S-2P0FTS$P?!MF!H4{#QuX4khiLjcYl$TheER9qPgV|#P
zD!#3SGvlAL#BR&!Y4l*|mL@XiE^Zil!4*{afMTvo;O}$@%ku(R&^mkJ3JYZw1G0?@
zXSYap`<)Yf`)+jJ!i4NQt>JP8TV>L2ozZRi7#J35V5sFHq`1c+XksEcP8WW<&4uip
z;Tk6SZOO|IyyQkY`%z<%1B+f=r7c=*^nP(YhAw*odADSwwaQDYi(hc5M4oLOHIWqL
zwqfDj!SsIfY1xZ-Jvt{j508GsxlWZU^ZI#>&PMyP`Hs4n$`)aDfe+4%iX^jfCwcUP
zdHATMicg;=Qtf{Vl=dZoH7p3gnghh_W}V>fgS*rA&bjQh)lC||%Ni}ae!^#?Dl7lk
z6~o6K<I_5y5zJ>r=pPnt)ciSYP>MQMxD~PYDF)<xEQPzYaF(4Ci%W)`MK>alIn0fu
ziHnY*{$V2JjQGKAH(#I(3sv$#0R7B1g;{(!nGX?uj$VccD)Z+4l>roJGDP&Kt^k)^
zW!iB!S!AFC9&I~CFW#$2*T0*?`#0$FUl!gpIzE**KTF1sMmfs5J(^55HX|w~pJ75f
z>8o1Mxm(BS&=qH7w~j>U-iPEQSI3<iL%HUTD2cSOni|98vHH((ynGZxq2F>@kIUU4
z>r}!H^!C8jui28(rxm50?i5K@P9+#!ctPLxbmQ#!o!rVE5`Hq#ikfcTK;6L4l3nsS
z&^E7=Et_(Y?HyMuJKMzoDhCRvuX-F;^+|)zZxg)VWI^4-F4M%2BI>zNJ=f^v8VoMb
z;+D3*;rq6M`^IVsKbtaX1xm0*bQrYt6KM59f2e4kl|<WQ(}EX4@YHPO+O5&Fbj<)v
zk^hfWsfJ{$3UII`P%?h5p+xiEFB}=sz*_p-QSy?#*k}-n=x)Mu{nrLv=WN0C5Ro+x
zG>6%cb{;+C0SqRUFxd}Xk}rEFsm!@Xmq#nY(<_>}iA;3y$SgP*^rdJOBQnr?#>(f!
zVqCY2Z2Sui$=HG;=)1lf;wO6wj)fa`Y#fGqvsgN^(iH(CV`VLiu4Aoc0Q`q*)7Qu|
z!Vy|58$NOz`8Umi;)6PhDRSUHv=-C!lDY6I`wdUy$Lv;o5Tdr^Fn%@(Hy_xNu2~KY
zlWt-0TEQ{xpUTcw7xIx2u@bA={k)}0n@ojcWaklmS?j{7n5J@o&op%seb27w|0Rm1
ztS-gB%ztd8eznBTX%S_PJVUmfpJ1FM9!m2{c(piR)u!L2ug=w2P+*IU3?Ewe#~A9e
ztL(p3+bCjpoM2aF<zAST&l@75@XTA8d=mYsNHLH#+~`K5jmNVVaSl>$SWQbZeX-%T
z6=L%DBHyr+^k;j27V#`fJO}Lg#juxnsvg2`jr+*^_KCnNvw3`J|9fzr>nppi{#+ur
zauDUuuAwDa&tw%}dy%@%9XfSt47r}(n49aG#ZrCR`I+*QG%qxd?F=-a13h=5`hkwv
zQ|zFW?>z*g-3xQ%pP=Yim}H^#I9zng##jF^Ebo>H_3~miLstQ#zJ;*!DyK+W=Oiii
zoPhkZMLCb=ZHMs+)7+>#iewO8M&I6j!iO3I94K)|rEp>1yYrP@QShaW8~y3h-HQ^3
zq-`uJ{~y19`zop{2hxVT!IXULFn(|RjlJsoaN&d-n|Jo3?AY;8S{q_R&Z!fSqh>~G
zI=N!k{{)48<I%m~59>$|q9dzE;PB)i`gHM?<kI+F6mat{KK8oL8`h^n$EUOO#l04!
zEZl@*=lzJSodw5jS$Ng*ToQfY2=Wt0kkSoZIxQP5y7fv}*b*%=1@-WH_Y~86Wsq#B
zj4x~*L{9U!;$^%x2JJP1?au8`s0n5B)V0vzDzZb@!`ad=?xHhw07f@Gsrc1W>d`9&
zJ*^^fVqq;ts3ag<+#UK>HS>0vEq>^3!|vIS*oifZabCTiueQvhgvv1Tm}-WNuMX0q
zDYZs1qH~aY;U*sL8HI^rU*w;m!Jwr^Q}0c~pb7`{J@gqR&q8QI!X|tf+k?f##z7L^
zB>T89f_8RMXC+FbWXGQ8u)Bd(aA7ORCR|2s>*Pso?_E4vwJ=w$PH-$Ie32QAPNOID
zQe@+GhvC}!!?51yf#?(BETeUqd2R}#zMo&w<{MY2{oEDweH@otv;HCye%c~n`T}ay
zQ<FI6r_!g<{m63%@lTGS*r#ZTUh18Z{^JI&EDn`riR_?t;RW7mo{FIMW9Yi)vh2*l
zemEaoD(Mh=j;V)Tabm+K?CSkmrZWDJ?182)MWr0Yr_54(Ip{*eHwB}5?^7NcD0pj@
zz1Zz<@vN`NN}FWw;Dc7KgR{kRnPU%ENrxVg`Zp2rNyFhOca=0X4^W@a@>sV)c+1;;
zWur$uV>50VV$(V^3UoV&Fmtg-M6)c~RD54gC<uN?2_N`d<c{`FC5QGTtg<~!gP9zP
zyav#Yo1wIIus6Ew>EI?iLovIb93LILl6H^Hmu&d$OqzpKDel8@<PTVkabKgz{z5um
z`80+MBlFpQt8-!wD>$ximSUlF9x8eUqjO`R%&fzoX9iotIerIX7Fc37+lB7455aS(
z=<{pn;Nohc$3ZFB(pra2?i<kU;3af@7(>`JNcK2KFw|%l6nF1~et8_^Do@bNJ>!t7
zcZb_`e}FCT7Sgckq6e(|7Y5Fk$hY}5I(Ju=-mvzkusLb4-Su6}=qu6YZceHi190U1
zH46F~$9>PeWX_uXNN(lpoK;We(8|iKtSb8>t`;R=$!iaE6`4h8l^BGdO~wt6bUwA>
zAU*F;q20csyRrEVcC%R+>F<f79(jbx*)-+93cf`Dj?8X-F%5FLCXuRs$Hr!FN#>cW
z{LvE+-a6qQGxurc`z|I)v}?ntx0ef4s{Pr*jkl<H$S8OQZDY&BV&Hr4hRoRf40COb
z#ruAX`Npfg*o8^`s9^tfauQv>_UJmvrHy{5m?uy3y9gfukEam}8ZbSsgem<DMZz&R
zI@ZI8Gz#4%DKBG@>G*)Y{P7^IB@y`V*Hx5{isYH~s`TxyV8WPhLnpz)+OKkpdD^@H
zER0!8^hXMGSVLx)uVd#o4N0-*e8k?n!amJ9hNj)8sMYWoY>)a2W==EP8U0$erZk-O
zHt0zy8tEA2Hj&y3rt$f;Ygyo@ZkYHck-wOk3XAWK*k|{Evr$v|@d2H1erc1)!#7~0
zlS*E`ID6#SA3#k11vow>2tQNyvbGl@qcQ20WVKH+uFPtq7XL6hx;<0yYUbi?Xdbl1
zou$}n0#tQ-P@C#J`qH_F^v(IHWOM&IB$=V;QR#}Drx$Q<c^>P1<2PT|BgxRV>HnUJ
zPPpb%g*qEQbd;4)>wrIa;TVG{okwB2-!ooj9Y#7`J5g@+2P!|^gSY)8OgJ5i<g@Fs
z>3lNZZg&nFgO9K-%WCj;nioZXam9%65wbT%=9s$i61OwBMZVuwBBwQ&Zd)v4R!*bX
z-K@8CdY(Of`FtA@<xP}RBC^T#eaJ8JJO6Rq74g>;sNz;EdlT-DSAVW?gDzee>F7g8
z9+PaAFK1qZylCHS2TC=1&Rk>8BInCkcJSkLiA~{i4Eu0|N~7brNxnEgg{#rX7g;#G
zQIkD5U5?(1?@A8)J;HR6`PF-Pm7QJKm7csi%O`84kda;yQjhkcU%KjS+pZKixX4rd
z@lhB)sf(=F{cFff^hIKyViZTDLwU+inVgRd*9B`^^-3GtwMK{RUiaqm%7LOUDv$Ns
zV)>aB;Y@RH29gt|z<A+$d`-)uZC2M=&-oSjb1h4p{r&O2Jf56xR<dWYFX+|!V60W>
zEX`L_6y6vwysFaTXJ3S%HL-(d8P>85-s$vDNB9bLR$__08LVP;pz$MwUDzB$bCR!!
z&x4lHi0@<R{bWPTSfqgc8c*qQM=W3c%nQn2yFzjMLri!n_%Y_0G<{4aJh!&8<Rmw2
z4~vD_$U*qz;fc5jhbdk|lb+g_)B88`ko)`&qP9Nd^&cfPoc*DOxkR-lb_CUH7<rqs
zr}pZk-KB_7S4+Mb<Iyujl{yAgAmr#xe)`FJ=Df?1njI4aXD<pjQeBw0i6KS)3T1Mq
zS7WQiK3X95aVHw%F~9gaNs`;>)>tL!rlLCbQDGpKbeuwEj1{+9>mYvLC!m+Sl>EQv
z;{3|vs2kZ&R<&g%DOHBDr&VrHTmDj}^ERA3$C*&?AVb{n>z3!Qqy{<RkR9=0K8{VQ
z<?dbUu_@XdJ2o>avIs)k^$_R{0&U*j1C=+dsV%6tG<(}BX3_Z=&g==p;Z{GSEQl1n
zBPq_dxQl#CB2Dwk<f@r&a2`<07VM8j`ieb>>2RgH&vlvYZdc(uybF~Tb7{SnoOGOI
z8j5E%@=r}W&}Vf!?`pA%eyk0r#;*6U>F6-gFO{%8>kpI0?Jumanj!Qg6QS*zPEOup
zA3s_?ucK=O-ljB>V~-@+Wo1*In?4cwfA5k0OI@zlZb^|=U)XE=@3O>Sf7!C&MQ9oC
z$^xvONSf}pz`W@%R95-W=r}ua&#aQ9%f4{_b{=12cL+0fY^DC=ck{e9B??((iXeUW
z-0(LyC5N{sk=(bHvhGK+Xr;+Qv>i>N`UM4&0`CKS+g1+>Up<eKwB_=a%$tgv=EAjE
zJQ<$S09-m{j!P@gFvtEquy5gitYe-ykGIB)#QJ~ame4hC7e9f@za!z=Z$a**?C-?N
z*TC$=bEw;RB2>8>CH8vCan&D67T;Kj$iBXe9f4n-O~~32%qBG*V-BmXQJ(oH#2xp;
z<uM;*NB%7n_q2bgSv*U4`!cAie+;*4$)mARnPNtIiMq(IqlT6ynf*_}qe{z%x1kj+
zN-cxapq{M6S04MORI+~>`sCavk#@~Ghy0R<)U7g*`t99G_j^8)JiQvqhL`lo_1daH
z2R2WDZ({>)d^iN<x*81n_!}$4=Q(4q8f~~DSmskIu<=x1n8!3j^@SAsg=g?Zd_7+{
zTAkKLDT?>*X^h%_hjP1F(-PxY?i<;Nx-ln6Sv7&xy+43SOhwEU>&Q32-Y7Zb1}<&i
zNftATuyV-<ib=~s&;D{$m2i+U9~n~rJOwhYKhAPyg)jxPc($a|Ho+8B!Ai?lY=Le8
zdysJrW-k}=jhS&wwK|6`&F&7a;)Ys_A($9Gf+7O$%g$SxFx_$&KJ(B>gc)T+Ud$yP
ze+!|my(iP~!5{FeUX9M5PUaVly3&;M$C#W>ocKQ-m;4;pghw6k@k^q}YCV6kz4sU3
zn&8Zqp6DW-U^@+yl=P_m%4+V^kWWKrt>F*eT|vmiFL>3lU+m+P=;5lB{7knaJnO&R
zC{uV!;=oD^bhl&vf%WLr?H(IuACCspu6eyS-@wW8H`vp7ft$2`<$7n|2sd&E6rCc;
zZs8H^cX2{S<|7vR@g!>9+6?b3KglltJSX_iG2#pzj+p&Ps1%=>Qf-l$bBX2u{d7Q#
z;|-$^Y&h%w(vP`$nX=(M>m@_>52b$RKO;lo8jG<QP8ZyQD9&mpY5oht%wZ{5(XR?#
zKjft0YF^APG?uQl^hTv#8k}-0Y2>@R?8V6&SkXnWWO^NDGrq>db5$hTOh1#O!4@8W
zdI9OC_QJRY3n)88nJP<N(73D>bN%vgKgt>Kc?kJ_p6C?2h>H8pA_bAfY^Xm+X3JY~
zPotS+j=7k&H%sIIXUPg`jj?dqVsvyJL0&P=?6R0EL`D?ic}5^?-rc9d=D&QE!bkeP
z(T^TJkigh>2VV4B$NTgY^O(Mg!ug?&AH(Hvdh{&es~do1rk3|)hZ{bs6vE(>2X%cX
z+(NpgI6W+!oa)T*ZiNVi=FR2$3koUm^({%l=1Q@%DWE4}r}An}Fpac4h~LXqVg5Rw
zKFz4a!Pzw;gJ6yCKTguW9o@OdfebpeGn{rGyhuHiv?w$D4o$s!ob2yT<y)2<q^`!z
zFm1iXPBs!HbUsUG`&bEPg*9GWI)&`RE$rws!LifzVV*A%Ffm_I^m=yClr#3IG9CnD
zrBUp?SsiZn>mzA>Sxz(F9-^eaU0C_IKr&c91p|M-msoB%%36F*W9gbPn8SxdJ4<AT
z+lt^{#b985jPE^sMp9Di$#ujm!L|1s!+o2c)1tGvJaM~^WVB=woYLKSn1c)N>#|MK
zRZ>HmV#b+}YK$RuN=)vx0*%e@LNiR8Xozwht1Wh>=tZj#t@e+vjeLlMohL%}@I1eH
zW*}>OH3tg}toZo0U4pk8f&}?Se&f2}p(-cjYKIH&x!6S%=p4q*0!=tRm`T+erlV?6
zB^lhjjyc{_@L<U)_{PXVsd^NwmI=mV)gS)u*i_h*8Bp!d7`CfrA#>g$VDu|bK>np0
z4b+=MQ{Q&xOC8S8kDob=ow<mZHT!c@Oh3Rb>n|pd7U?YeMa>(v(bu>L&58Tb^{53+
z`CW;uoBm*9pJSR`FaAt95p@Gbiq7amxbIqyZ<ih69}r9vlHBqCd0B7Q{y1iKi}YvJ
z!)!wZPOa#a=VqhL)Q4@PlFx5N7c><+pC8BC>B^Y)eJ?Df*_69>0_xqjN&3#7Os2iU
zaU(>V<_w%fKW_Jtz0&pLZ36<)5UhxfnIE}&<x+%3Z9&Du?M4Y^=I|^~WT7+UagWC`
z+p$|PAtnLSZhP>9^H<`>%3!oC?JP}}yfIp`ZVFmHJd%0*3B$oRBFA|>n7ODmb7m?}
zrhhwfef0X_d!7_UABLjmN-1|4P=rShCz8f3YwD+XfU2XSFsES`bX7%HW@!O4FAI}A
z|2386ZOj1KET^plgAh{Z%J&a>&J9|-^B$(9VoucpdG|=1NjIahK3`d0`eO3z<HL_A
zoX7izNg%ad^n3jX(b;;+4tk%VH)&JxY>&C@<IQ~Ww@wuEO2Kn#ilT%Ag*a$XiWeUb
z(dgHmq!Bm&lDDEbTfdc~?Yj){sj3NQwTtOlkMY!c-vUQOCTH!lQrU)Kw<LB~{f$Nx
z1whXz9S`coOzo)X%Fq2G^D4`w<0jXrz41Qw%^!>6s!fRB+lhkaykZC49<$*q=H(u?
z4W`;b&uOV`x|pvi@st!L7;f)|#KKF6QQgZD@A$Aq-{N3s`i=VjInQDJhg*9;!kiDy
zxNi3r)z>DG?az3=L*WRzdb#ssC%k#&AzfDSI}HBM2cSB733=9L;@_ihxt*F~VJi@N
z9ox_1d_*wy*UF&`$AK6)r54uK{Xj=f8h!E*`_FMNxXg45{WG{DSVBc)BUj4|&4ZcE
zl@^L?+JvQVGuX({u257{l_s_>peM3g#;sqH=JT^?{_7=j6h~y2tUiiP*DFj&C}2Ze
z=I}?uQ`t5HJ!tJyWDlpB@LO~hCbjb=#~v?&iFFp9SrrLhm?FBqsT8>_;g+=DPn{Ly
zq=q)Hx%tm!l%RA5QAOK%%o7EvW2g-U{f)$bnPD&u48}9#R6ZhE?6_Z#!g8_q9<^pV
zhSCI)r#EKG%_V5hI-mQY(FPl+R?_`X8Up5ALQB&mJWKMXsNub6nw1i@S~pX)RExXG
zU!$hPY`hmcx&y<rQDMN0tR7Eeis$>GOnDQPq-8VLEi1UFFY-N~e)GJZ^3w8~ec`h4
zB)60{@O0B*+`3whj*J>kqh`%Sar6mH%XF9M*vz5fWiIsPwK*LSS$(<pPgs)s9a<Zh
zhck+Mp|E~6>204Z%eY&Lw^Mgwj6;8_{+S1bJ0JMs@>1qj*PBA>l=*tA<02y)Lz|Yr
z!?X0`?B9$vELbZE9bJr(^hpg%q+t|4?;!M3e0evuN5ndJ$0f&yyxlw-Dhoa$$D{|A
zi9Eo)jChQ<(!j0fp-kV_8*@L-0ynvhrDadqW9t#5*QtR?0+%3a{WDaKIw^c?UnQ@|
zh@G)?6t1_yvV-FN6E(3jjV~&IWfvtns$@pJ`@V;f><%_5o#Usg{$b(%t*~kfXNx>V
z*K7G1ruto<0-dv&_m0~*-Qtb0p)oL?I?-tOTSrQ?a-d%y`k*ecI#;dm82LT+L_}^R
zH<+p~oVN<RLu5Dle+<QxfC@?9*v>p(%z2iyWTAP@GJ2Ns5YyIA!`r+bd0R(}rGs1U
z@fSM|(81~sKBH+R_D{FK>X#u{bLA-dt{w{GeY+^RZ7k0G7eLu7<7i~-6!^N%V#h|E
z<BONyf!s{dzjl_F9yC#vCfxU<&oenWn0}$1w%O?R!;VLuBGk-(441*Bbgw)Q9p5Xt
z|DQO#u(}SbTM4jP-vQaU21>o|C9~Yp%%W=t!pd$bzwjp*>fP2*-q$*2c>NCETy>=S
z2b-xjWC#XrKZp${tFYxnE^N9saXG7uY~T7Luy_6nbh?1S{ko94qHwx}oM0>bPO*3u
zJ&`$S$IbQu)W>iV4H-NiTOa13=%N*+HZ7nw{ZVjOR6y(OK2t=phv4L_gT<%bv{LJ;
z$RCx!Yikh&RC$womv21acr*TVi@>Kr=V_F!GfsJLg4}vdF>grVdTaY2W$Ztc&q?9a
zK6qmKJbSjzV|(tn$%`SUQId7=L^K>1C@}vX+Oj=Cq7yr%06zZ-mr2lgTpt-mBmB(Z
z>OYL1{HRT{wpY;V%<kBbdXO4jqEH;*$7h@s4#QoyB*8zYGP#LjpD^?`AAIs8w~eZ1
z)eqgUHSG>w&y>rHT)CCB=Pl*Cj`!x5#?;~1&S7NntD5!BcuhWKaY(oL!K{Yumsxjv
z1gDcR^fYo8$>vJMd*mE!Ug%NyemCmBx`XA8R^<*unkdca9&{c}A+Nv9tfa;Z>SL{u
z7+u7>ER>h7D4)b0tnUoIyc$OOU%1@!yGGl)q|hzh7Sx&eK;KpHHurs`Gt>{#Tr<h0
zK51}P+k{a0Dn2|g9T{0kc=X&bXIAKA($*e;`lEBWWyt>Ae!4-jtqGp!dZj}2uDsY+
zvqZXkdJiqG5cA~+r&!OC`gC;gFXoj~ExR=-RMPnOG}Z42#-nRHX^Q#|v9q~C0V>IG
zirzyz-5${&Nguv@h=e3LiJ0`S8dFxc;m6=vq~A3dL)P4odBzp<`zu5rYQ|{{wRK~2
z$JDSp-TP9bR$q$o5)SHZRbnrDUUU>*^P4?nSSY$Go#xn+-L_9qx@(BUOD^QUPH+|1
zR3LO}Df@3&90iYkgtI$M+0M0TsQr||8mFJ4j+_9xeLPt5<Y+JcJlX?&G(xc8v2YhZ
zu%{!hZ@~D*G^56JZ*ebvO)=KDX@kUvDlB7J{47<PHs=w0a5WD7558qqjfY_K^%XjL
zD%0V|q2hJ4TNd1UjIzz9lJDpUGMG4-hP^)vrHrw(LZ--Tzb4?a?G~In+esSUkb>Zh
z&G5T14BmPF(I$_1crpAI-?;m+r1z@hEPBc%a%;SXg$okV{nh*2K7B;D-}Vev>GhSW
zU#Vf`^p8C%{=rt<{my;XJ>W}rcfzdIcleE7t@Jro0cvU~7&W#R`}uRDk)HoDJY6<I
zbW74jKgojLT^kOSU=2*mn?<Hed(+KN=6p-xZ_1jRC~H_L=3K*nvK<F*(47~)tiw76
z)x~1ZK5_!ZY<*<p;r^X^7>2W<g4JQ5sKt9FrP1wdH$LXU9V&YkMUB5-B1SL;mp{pa
z@vvb+u(OJiHW%XS?fsbk<^p&7z7|tVl=EhGSx+7z&gkR-EVj6Y#3xdmbQ=lXo448A
zrt$2Jx;pF|PhyqfS=98JfqkzFS&>bgZ149^TwCV{TePcA@@iTv&Fdb9(}Kr#uxAQW
zGKpZMc^8AUFXO&+Kcq1+_$;z80UhzE6&}~$N|}^rk|fz<)f1Kv*Mr>LcqgCUsH<u~
z?I~4Ccwj>V9A==f({Sut{)sg!rsBG}O5XNS2jOw;2x+Wcz@`>=V$Ivq`JW-FEUng&
z9$kMx89P@)x=5AAkDV+j&7X_>%a1`Hnz>x>Za89c6W3}-FfZXCuJ-KC&HoLD>Wcdu
z(%t0Q><^QxDOgp&>7q+I+}^sd=AYxZ?eVwtcZRp{%EnXq-ofZ$p#i-Wk<e>efaHy-
z7*VqsKXUJKS&|ZrOS5<*<)P$k2y@vQf+M>P$zZo9ifh|&$wYWwlf?H~=^MY@B@ik)
zro64Y1*B=Iw343k)Fwr#r+o=6imQR**f=C99OEs~N3n3<41AuV&+Q9(O6^u}pu)Ov
ztl2pfrb`ti9rw3OI-E?PYJY)07-LJDuAX9_eiX7JC*MKaWgngf#bS`BnBUH>W#0?D
zuzLLwc0^5hvFk6PgvY@<dmuJvd_qNd7-qT8#H)diar2+M$kA%zOq!xJ?)pHS7R;NN
zubM1selFLmJSVwyI}iIFEXBzCa*}4P9JXtWH(mC<j_G0#EH^kIYqR7&JSPj*@UQ#W
z-suj)>qhby?{iRb;3oXmFUNQP30yT{FrE+3<$2vN$j01$LUv;}vbCLBd6Kbm-q@Xa
ztm@Dcj5@Ut`hM+_O(S9uKIH{Wz70dvrBBS`hX-@8Oyo<q*P!J_Z`wb(J0i`dOOj>H
zZ1YtGzMY-KQvQ^-DxE^=swPRl-OkK>YAjUTns|_3ENk~Z#0{^G6`r!qP;X12-941^
zPK8wQHO~cya%UR;(<@=K=cMC8g%`}dGpWzzOH^`e5bY_|rI)%tBvF={baMI|Y!dsH
zEBhU2XL@&XAEySo9g0&u6C@2sN|<qQ9KIB8VGk1XF!IA}diSRs+WUeraNj++uYJfG
z4gIm;`Z+rPi8H4UQ|R<<U1UbQ#vZ}$y4<`JiX-*VO+E}2n#CBMlSXYS7ZEi~j>=!m
zhs#MBbOsvm)W|#RtA(P}#qK$M*&#ZxZ?wtqO*kE$l_hJh-b%KU+(~(dBC2-<p?rov
zcKr$C7SB>7G4CTVYi>Dhuepf9+Xp}rX-FndPKlo5dDLp%r@P{NF#kUVlqWdhQT#7%
zIsHA=|MNBK9cTmJy2BES^D*??I+3X#&thza7B}2!&zv5`@r~`#bS@zcW}Dw5cgZ2i
z*lR{K)#Dqt3-+Z6%P*lIcoY75&lbK(b?7P=u}ZIr&^H*4#}D<<w$BOQqVf<K+Ku80
z(@3yNk(f}7HEG@H_Vx_+$h!uI(i$+~mK;(-HzR1bK0kM+2s=w1vF^uHR_$|+SM2-_
z6{ig7?i1|MPnAs0dtA1P=oQJ-&+;#SL=QmiEjw87RAT$43~To6z}NN+oKdJCo#Fqm
z$#pa`St0@l>N91rJ9Q9jjnB?g*#++~$#D1ixr4^cLHhektf7A<`tItBxL?s!c=sIZ
zsqu>IjurP_rNdCZ*#mVw@?@4TyX1ORoT9_^)-d%@pjRfr=>K{VrB3dYw{yT-!Bt+z
zejar}qU<zH`7EEe<Xj0ZK2act_lK$X?kt*-TgIF^{bjY14E}tna^A?JmGm#F3^!jF
z<3y;)Ks#u_No^V1U40hGi+*90g%gVIsNj-IDSNK55q4$wY0T(o@%O8eHE-C-#&5ij
ze7S7wbaE%9ol7D8ycH&vyAj-{C#m!8WN3ICu3gh)smtxrv~N4<ntKR8%sG@q1Pfp1
zA4*L9g@m$XtonG0dU#&u$@g`sJ?=N?PYj>+@;_>AeTbfx`?%djPpq5XRXY0OL$+hn
zXu9Aj$3HB8h?)QOpz2X?C0_)0)?{)tN*}K!tAuiV94bL%P$Fh@DrJSAa;PnEJQCJh
zv(Ev8X<N-1x^6cK_x@hQm5>iI`Kn6cz!(qhlY8m!n>47aOk|^<>QQ=7B-@<fhj($=
zusB^pUI%2tF=R`M?cVgDM>Umgp2UBZA4XlX2TpFTkW6&5q+L2+(f98{GI6ki{mXRO
z<l6@!-6i_p%Sy2B`dmrt*f@&OP?j!_EM%RI{lZr3E3_hDChWFWFw>>Fn7HqSWKyF_
zo?h$<)_O||^)`ZU5PpX3`{PabN}QRy)jminFTyAz1f%A;v8Ey)QXlUHr|oZG@wZhn
zYKJYA_FssikmsnE)}e5FF8vsw!CdB+QN%bWDu@q(p^ae7^=+mDtvktgkB_Wvkvt~b
zx3M#VQSxzcE^77<Vz;|45j#>FTC(yaT!q`GGIJPhUMY6{Kfbf*kDge4WdM!1B0QPX
zu5t71?~-MwQ|a+ITa2wN<Bz@w_VK`2SoH2JHE*kAwL_2Kfvz%6+q_5W+DKYG?<_73
z_m&jj{vxTHatp`ZI?<DLJusuIj_7!u#f=S)Wa%=Vd3=)MQiKPTZ@<Rf9Ya|41~u_K
z8$$c`4&fdo$7&ruN_vno?4LW}S<OgFU#g6(6PIvczAN<F<Kf)8i5$D7;;K8RQItgI
z7C*(%h$Km8hici&jZgRojd02`){yNQ(tuevn&|!ED#^qpk$iGzCq8#_Z>)Lq5I@3)
zvq$l9Oml1%O+Qn~K8oj>+EPF9{*=<+z2X@-JOFYd(@|;O0aw3y)S#%C*Q);kho7&e
z95-uRUzkZf`}okR51+6i`Z(ET{zuc!1+YgukHY3c1ll(%NcB#i!H7^3BnT$>)Tzk`
z%@$1NL60bOOg6Tw>qs`0%p&zgHT>Y<3f^LO#3-e)3=RD{NqbpNp|$0mVbdB1SHTht
z@R*3u+X{IjdwA1)SA{&2>L5%mIF6Vuodmx$owV&2u}d$`V!-)%n47jn;;7?~^*#|)
zx4IqiceldKrVwr;@6(do9ys_y%onPr@v|$8DaqCfgX6km?z&Yp(p^bfGESd57WKh;
z&)2Z8e1X);tAcCdDa)VriMEfvN)uM=vbrfB@X|1amgk;iHrtf*w9X3^;h`k{ZXxh2
z?3g5|tpuCqhCrj@3L;Erz@zUGa#lCMAj3WMSLApdE)OTm8}4-O$8s9I=@+gYbLJ<F
z%OpcS9Av)+cBbh0W03IbqpY>ffwT-_Sj_3=lJ>z%C@16rTh@KNWJ=Zpatc4nBR*I1
zj8T<z)AkVdyG{q~2}S56)jWNhW%#Z-n?@%`a+9QD(lp9Icbz7_Wz~5q+f~Yr$|_)H
z;m!kV-r`7DU$Qx|iuqj5;_v6A(K<yRl8aE`gLJ&0Uz8-X-5<@*mnHKn<0iA+{R-I1
z-ji_2JPxT|rw}k?s7&WmDSuGcBH7j^9BaYKQj?q%lm_%9bCqZsJLV=v-!Fyoyv<nm
z$B_Dcnan(&iJaY#4|vedka1aoR(;MvrG8IZ*&xGedtW-a+8>vO=doGsv3NBqpY=X9
zQS4$b5C;4pUo*kBIkuYHT^BpwbYSJ|SS(YVCi4C_`A=2R0c^j{dp^8HJwHx{#!#Mn
zXM6y^+gHq$^dr&zMl!$JKa5REp3aTTdrGbQZDv=#=rhfX8g#y%E(zU5xZcyB#y`pA
z0g?XbksQIxmZmYKvFAzW#7|ThUdQYpC8S)BLUPMIc1nIdig&hRnSKDyy}yjg2VK}V
zv4hdP{*R^Z=q1&u)-yW)DTuOclQBnGf$l7GMu_?f9Bgc1tL8uBN=mw<pS+$;dV7Vn
zFL}jTawl8}Tm#FD<rL}qouAtFj`y0APS+O9Mzz%+{`~z_+A_eG+bueQTRGn3mb8Qi
zyz*gB{Jl}E8_qXf=uKPNmIww~9ZmJ_M4xWAuz<<^aU*;KtvK+HA?=gs$KS>lgF)Q-
zcOJF5{zr;uu2R)oHJCQ8r+|Vm!Hq4KXeAo*ZQ1SYVeLYi)yag?@)nV4VKV<c^*Db0
zUWxLn*)SP>h;%&!yXK2O4vj0qkiCD{nxru-VZ|BDe`O$1-&}=w$NSh{5JJmebmql!
z+UT0)2ow3K_}8EYdqXSydUaX0H&4(^z9jH&H%3y+e~aKeHVCf<gz;VmAIWv@0@gmP
zR5q!)i52d@NT(Oo(4@s<u_n}ooUYtQeYqAlPj1JWw&PSKd^xJMPkG^_Q}FM0lw{kB
zsoUYMke}*BK_i@@T%U*`>&BwM(BDWu&H~~S#Uf5NQR5a{7JfoZMEhKU@riI*n&UL)
z9<E68A_ucA!i5KK2!nM`E!rve%Wl#6v~9jVKmR3~Jj(RY_tJ1#d3HG04Sqt4JAdHO
zGg^79PZ;|4D@06vAqte%;lzJwP|BJ{dz43mr-z|nL@}oSJdd6QF*u=RM00MxXCs>u
z(LX|l*v1NW*}WJmEbg-Ml3UocAz0Gw?FQ<v@eSv<)xg|7o1V0QhPwqA`B;{+a}#gU
z%h-$j%glMe#0~U&kul6BR5L0R@5~D?Avg5Nj2mBMEOhHcnARC#&(O1yu!g}b`tDFl
ztuVnB6J_{xoS|t+;u&w&fq(&wR%?!+X5R(`y*NPAOdRNAc|6wM>&_+jx{?3O7CuP&
z3HIkr@u}h|Ii?6Q-N3C}{az~u1jtLHHYT%9Nk6gZjuYR0wV8%G&&9?5&q+gdC=GwT
z9EwdE{AHnV8fk|!mo&j>XgUESvvl<Slf@?Xdr2Mpb>LdY;deg*%8JWzbNp-e&EN(;
zZW>O8{ney%d?noMP&r(#E5NsU6#@>P<)0_5hE7Q)Y_3mc#t!FM%ezN#`Qb;WA0;5&
z!BOU|puuJRuEOBmd6KNZ$)avXvSQ6lo@sv?l6hBnsa^*~kJ=*c^WAuYd^CmG^r0tT
z&%rD#fmbEYrPJ=c`E*Thn(-=xn8N_beO6;J>yk10P$c^k{1DNT`pSlO_+#+q(OB~1
z3k+?)usdCSn2we{#o7L57Yh`n2lK!#P5&ymV%Nzh<D+cWXHAN&?S}<DJy1I2JQ{z?
zNnY+Kq-puFm~^v(4c;U6nhz?-rOz-5H;Bc3^Bh<{HOK3eQ)n-~$8>V7jl#Z_@woxT
zsFDTH^5<9C*lXtz-0wdcJja1)KeXj-ofgx7vmI%5iRe{me#V)Ty)o8mt%Og!f{e|c
z_>dMQoY6&4?U99^`P#JjuqD);0+3l>Lt$_4OTIK~Q|=;;db=I0>T(V=t*;2)pSvhj
z%z|d(H}1afqHOXyE4nx(RN}K!0m0^zC@MRg9r^2rQkP}W;7zD4u4H|#jlqVJKA3x<
z0>k4T;#X4%X*uNM_l0<o%dtm9(p_#ecPItk?gt}X3sMXX0hvC->wAh&3svCtM^o5m
z8+R<ax*u;;Z=(O9X5^Bd<b%de_#T)|))9RWAzzIntx8f8cd>WzYlr;wGpJMagv}lU
zIvf*7jyumGEnyX1yf{hhjm0yhVlrigO`&*w3*?MHO}@YK`K}^uD2qPc%q!(|X@?;k
z?R@d*$4t_ElZ?6L?XoweI_Np%B04PYv1ykBP_<?(Ja>1JqW&sbPWhCZdbbB+td~lD
z)omfKV_Rhrb;3!#!Ii&CT}-dSt7QA%$mP9Pd5MaTn^|r}4Kj<|pj2yzlGb#(+Axj>
z{=EX()l+=rop%_PT8NC7VSMH3NhDFKp*Bxnvhxk1YwF*T*Hc0I?(GFkocw{wx);D-
zYDycIo#b*pU307kPiIX7oslK-s$bWvhU?8JRx!SmO1!#C4+|FCMExR^nVewFQ?`TT
zYtUyxh`1-7rze>YaN$ybB;Cs&d%he)-a;*SoVF9ajcYhJv>uah6{6_1AHPx^0Qs<=
zG_*8Vwk!J_KWS<&_&OqQ5I7hO|3nvI;6HZG|1ob{e_pa=n3CYdogi616MFVKi}Y#_
z<F8nAEjf3OAD^`owK+j-xXDWLJ=KQ&W;!gT={)PL@m=Kq@AB#qZM?H2l#O)BBEy<-
zP^@Yd9sJuE*58+{+jA**JpuJk^(p1tdlqSFz@08^WJ@Mmi?it!n$-Id7Hmp{UdKx|
z>FGpX{Cz4`t<J-pM@o6Sdwi7WHwgdm+@&Phanq>A<OF)$dyXex>!4yQSRoDmq+C9f
z^;+A(ht_wJmh4iKK3OyhQ^fCsr3cf$-s-4-+mAH+R-!TTFX&q<+jKD=a&kSnmbh2y
z6-;2BH~L|wyFTRX7vkWGC<JNr!LBuq41piHT^B#2!>=OHOXn5ytu6%GEfL&ti5)BW
zEPMI7KS@(B(G2hRP&pAwGlno)^4o-N*Pn)U<3>z7r3rJVcJ_YmAo@DJ7e*(`v7D~%
zw99ueet!(43y*d2?4m6#dpzH$?bB__>iyT~=$a7LUz@Q(_iv!zOwrT&a}n9&r_sy)
z9lYP9a+E53WEBfs_;{xf9NIdaENA)R$=NpS*3_rdZ%;_h)dVw@C+<`y=|jeMM#wsS
zxeDj$qxfEDkvF^bgfBZ@$A&JRLVZ4DVe&jndg30;C%N2){tM2x$Iqf!F7HU?uJ94J
z*Fh^<g}O9q(&7mt>B~R`=~$DqH05$3nkTM>(hX}mcxX1hExbZwrZiE|^v;O#dq4&_
zhL5Vl#lBz`E^A+fdSB3;PY+nykT%x1@dwlH@x*9rac782FIBV$!~EnbtaWvx#T$;Z
zA<uR}VNd{Ddh!(Io>9wd-{DFZo{4#Ed@u_sKTo5@Ot)bC3W<hPRhAN@CRO_x&F;U{
zL*>aw{6y(hSj_f9;TGXn2qh>ryP<m08N!0ave~0^xO-zOS1I3#jLB~$F+=lFx@;S^
zzkN!#g!i=}zdH>GwMB%_bmrc*mijC|N!9UTSaQToCadW~rSU)656^nmfAtfpV%hY#
zTE--HSLn=91+07LYE-eiE842ksQ=S?9;b7PVSs|*v@7S8?7B&dgJong_B8F|y>afd
ziC{?FVh_IV#%{S#3Zp=5$s8)!v9r+>(~fF`D!xp=2|lJTY54bFuyfFZ{ikR=>Xl>E
zx5a?(jqgNnHaDPs>0{V<`q785YZSeqjUDFl(xFw+kP2R;!|$_DePBR(_MOmoOQbCF
zh#T`7wajS6iA`v9d;-l$fg;yvjcwwdxKi;J?8aR)s+vEJHR@ks&KG7<(8XPNHMAA-
zs@i1v(+wwhI2*M>nQgF0l@(S6Gs%)Q$e%tRK~hEO%K@=?FytxkuMmbL9Thw;*^kDb
z67sDXMOxzPbNG~C3;S<@XW0UoUd|l2{Oryizq~@pQ?_75?IkFgji7x?E^uRKJs4<t
zpis3BNzLc-uRZ!go4!LMySMN{KSOH%5_bK$ljQjH`|wB+e#{$dF;lKDx>a|jf7!p8
zZ_TsZfSJeeU)vk;Z-*os5C6dEZ|7kD;|5>Xe{pVEaT(KV%cZX6ztBDUB;t~j$a+^E
zl@+d_;sGgae%V&+)VK|8tI4=2?ni244Pnz+h9N#Hp%{CSBu(|07y1%I*W6=;%X&(C
z%(+Vz>eWcSe@@o$suvwev!J2b*Vw&x3$XC$a%9AKvH4!p`Sw4?bkZV^(gr7?es$+O
zL(|U~QudDay$)f$*FQn#umDm%qC~^O6)0}U2X@-22zLJhL^mu9S@li1FPfTI_I^*Y
zmpN19-_8<?y~P+CF_5g~P3YixHB>$R%x9D<k)rWA%nd!p0xCpqZ(1_^JhDiC!yw-H
zFOoG6twGiNXkJ}dgf4#qft#Ah=<I~iV(umx|BiXzUrSrt522@vKQmk!LLohuAo;*D
z+1W7<pfSY~I<fgEc@fGVEjdf$Wh&BM$HcwHc?)7G9ZvZ(g}1mDbVQH2{);hg4m4+}
zzw)4Zb`=8q*f0%0Pq-%teulOwU+FP`8Wt-vOl!yCrMEGmqZpT}l%*xkS#0p{k+kK%
z4U{%Llj#l>p5B_Fvix<Ad4@ir{m**Dh@9})e<N|lCK+QKPw|`2x6z%V1YYtq7tRji
zGn<~rR~0|xmYFr+u>siQe~xQ^x<%6u9%6^h-twMPj|*Q_6kAjE7&~4^(1~;RF<#94
zluc4lF1s)GsP;Mg6%JBptG)1{zT+1UC`#W?t)@|<4l}QLwX%OdHjsz#+dp15nbhLU
z$hg3aFZy;%?3TKq+m-FG+FF8M?MdkS@er#S^P3Kz_(j=0Yq28lH9z?yA1;;^xN-R%
zpKbMkiiQcFrl#N{k1Sz{nv-eQ#8a~3zgcwca3Ea%C9xdg53^ED=Z;qQSU<NhsJZA$
zfgR7GSNWX9P71;1UdIvo?J^CM%ci<{D=_TFRNA?A0~^y=&epCzfiWMaQNy&qh*Ul%
zYqL$E!I1^1_TR=@A~bPrloHyHdB90I1KNu6QVrK#kZbB8995fnRk<4rJ+YtWnXV>f
z{}8m>d!t<NNj;ZFAns)%DwJ9!E8LW%rVaMkQffz|TgtE^GLDvv&ZI*T_o+6xTyUW;
zQqbK-xXwL>fd(6)J*ZT2P1c_h4@MwnUm44p^jfC9KMqr`JVpG8O|<avIfU%#i>BU6
zsQ388;tfx*k<SLR$}i_=OCy86cN%H?4WZKGi@3|QjmX(*hSlw>X>65Xxt9gg_%ADD
zjg#}SVwp2t-_gX2RI<@`&ofGx;ei=q|MYp6=vw^^6uF*s1a%q(Azo(26+VzQ6w>&o
zOEA-4PI~7-tZc^oRLQ!3E!<01j=RTWP}!V7&F7B7cg6q={NgTh>T~(x!`_m0e#Nj;
z-i6LD<lx_39~!s%;?T20)I>M%g3=^gJ|yvUR)edX!)5MQhafTVjBJ&|2Ac8mIMvC;
zaL0Fhk>Tji!%8clJShfFFBjvF(MZa7Y07=NdNz#GJgK3`3b(x7sK+ut&JyLMmOqMk
z&(J%hvgl6kA*mmQeLaj@Us@46J_@&1?8CyS7XHTbRqorF_nCuh9WDeovehFuk=nAW
zlxknapT3rd{?K@updE}3$t+yFe4iCZpMy^1Bw83Km<RG|tncx!JSnz6y%2N$3l0T%
z<Q7ZHBNA}RCK}4evQTF4hn-fd(Xst0Cyn`Nvt9}N<mt5IYbU{V6!%MCFRYyN2;=AO
z<1fXWY4^|npnb;_*H4zgt=~=>^W`}8t&q@mxs9+{7)nw5jp^zy6Z%f0NoR{Tp1#vT
z-2GP=+TcM?iyw<y$Qfw$afQ~PMp{#JmX0i*O3!C&O1$jUk+9i@8kK!1=<Hs~-+u|G
z2h2s+ifYMB%|7&XgFnXXU4w$iIE-nqqoKxb@Ux4-@-^o`WsfB)S(UVSM=-Cd7$XVI
z$)IPle@u4p3q`7^OLEHPq~B3S^4>Y<<K|0Y7EMN%_(*n0=_&gc`2@y?!=U!9h>!k$
z6`y@mNNYtlc^^NG;=jrWzB?XQ9`8o$pCjbjEs@48FeLRIWyotELA$E2(2Xt;!k65W
zPVCP^wb*A#Y&38(OP3{8X$o(eA7YD=DBMgPAs$|^I&3Aowf7e}jxEHZv@W=Ov4uG{
z3D>}XgIFJj25eaCBk~L7wED|F__T(ybrGt_u#Te8Ns)Nv??<N|n&ZuQce>FYN`2B*
z@;aGJBj4lasrl+oK7UCH2DZyl%?e%KyLlrr6CT6ceFo~McjvEW^g-f%MU*b;PDKaw
zki9LIHhFm=KIIt6>1H8v<~M$2ax$Jgk<YVP<H9muujF}iU!v1zThbD9`##JRR!1+Q
z=#(N<Vm64+X%X#FFT?)2t8~gI1?hIvXv)CBH2Wb#$F;$*{dt1s-}mFgEq7Ar=^12~
zYl{Z0?r;wFK=0#|aM;v|ylgG0zQLWZzvqRu%HnLd>myc}yW)HGW7hC`5xjDQ8%|@Q
zQR8DBDja!-wQBU?>vEN()rAAd)P5oz%K9x+jorvg*X7`YdOOCg`pc6?Wn;YAAaa)a
zOVs6CFmQdEtoK_b9Jq84@*k>^qM;~l)>WomHhaZcQyV`E9?SY|x*{3)=n}Fe)}(Rp
z4}!9lv2<1(qPI-IY{8pua68F9z7sqAnL03TNEP$zPH@_Cp5Kk!lIyzfIDa-g1^Q88
zJjUq@lo5ovfgeco@<++{xB^U;9EN+@1Z)uJ=W$B?Y2xJ?^3BmiezP9#lwY92{WZ+8
zPcw^&3WCGX8HnzsByAGeB>mD3*1R-@)wa2_j+QNwnOjHDgO1M9zfqsrp2iQXeUiuq
zMa<yaLo)Dkq6*APlkm-P1iss!Kz30w?O*JL+Y0p(+q({EynKrc>mSqihxx3VygQA#
zs?4uPd0<I>37a@dmtHg;<<FIZnZxlwOqntTixS6D;jsX+57NT=o$(@5b(VVi2gA9o
zlh}J)LHlkG@+%8ue*eYOn<qy|Grj>Q)+<WY6(+MF!x_lm@m{t}TZS`N0$@F;7Oi)E
zsQ0ta@VGsnB0c@EyFi!ysWeAU&oCP8S_N-OI*)R8mVC&sX0FRW8gcDhHguOS{oXg3
zW^cTQ`+4)pmEF$n`(z<SesY0xH&>*5oQ;U#kyscle(v-{e4o6BO*WMyqY3w5Vo@NI
z|JFs?`X?Oz!^~*$Bwb|P>x1Gyclp+c2DEHB1mm1c7%$1BmX!sPeTUj}$7va}&qec5
za#y%fQtqR`J_r8Us;H<NP05)*neiPpI<e6V4SOzN#-{7o99T+Y3!k9D*$?s^`{;bI
z35-@;79M&(_H5I2SwH0hGLjelY{wZGd815bKHU$>E4oO%UeBlVdMC-a>KHFK*Cfl>
zUo33$TD<=@nG&ySLg(~zczL|Y=@#WiFAw{nA-9y&mewL-SZ8Tie|eI>bCQz%9qEnJ
zbs9BEi=KV`A4BII&E*@0aWqJKD3$g=MOqTy=c1DOX=rN?X-OqXlkB~BHX)l3;``hp
z2?-G@DpZn|N=xha{@poFhxfei^W67!eLh(<{Js}PD_@|0^TM$qwu3J#Rpr-r=Mq-+
zBgs=vl=^awX!HwzrdHotcx9d)KVxtgv0=qjJvmbB-}D4E&Q4q@@HpxDjN_MU(=h#9
z8~qwK2*32=u<mO!sIH9r3GWq^Yv)ljRR$4;g6~e+mST*Gq2GUo;00gH4}Esyz3ap@
zEKi1g)NrK(N?BM`Fd5qo?Wr)m9BJ9NL_hX;i6l-uU<Erm*r1eaC<^{be+PAt(^D@f
zsgDqLQqS?UcLle5l!^|?!<e=09fHqwixUoR!}(9@*c0@c9b5K}r(b-5vAWHOS};I6
z-BVz0%@0DeUoP5v{n0qo57a-HKKQrMH|?ePZ*>V-ALxg?jb7rK#+f9F+CsjKPp~mw
z9lHiA(TA#T^b~yMO2RDBYK#N6iakXmjRWzzQN&i=TaLh0M$GAE1nd7qg?HDtU|s4z
z&Yr(R@~9YgeyJasCXPnPh6kcnuVqk;3u4w=cfcp;8!HdJ!+qwgE4;8vA8F1F?BTX2
z$Tu^f{HzBsG|wT+g+eCP)E&cW3WS324JObAQ1&7X$CWptPtg);w4BD5TKI9hye}BH
zYXn+4qUhibcQ)cjGFktU<fHm@z|U?5eO{$ZBNk<dyyFL9)S59+*z!c^Sj=L7XJ>Nj
z<7s^Hu^I#qS%n@QY3&gkCDGV$9Ahg6u=1$qaC>@B%kaq)79O}1hfWFJ`Qk=wp6Ngj
zRVGl#IVa7MCrk0=@_v!+4k_(^C8}iOxQXxBA+4QnH35Ai!=UiDfM*95;J3^^%2bUk
zSX3>rUfu39bCWjgdvFL+igS^7*_TS!R?w-w=Sb;e8HMV876<$lI`ir)aCwb_@JvRc
zvN?o3o-_$tCgmU|_#x@pT|nK$SPY3w5_R@%$M5RnRNBXcEip00NDJY4{+Wh&w>F%6
z>w--K#^7Lp9UXa;jAZZ2EZ<WC0~-8bc1DT66!y0-w;1xxG2=zwI(%{5_5ghvZ9@_%
zQA=)pG^N-uM{%T9D8Ds$K0k^zbR+mVn=+;!;^p$#7*l^N4;iU#J?Xx{=xHQ{EBD!|
zH-+?M*g#UYy^XZMC^*+UBR=K|#UA41RD7EM&76i8*~<~|>AkphpN>|1)Mu{iX@RPE
zLv*Y8i*>JzV&D52LH<oA)ZJh6%t!Hbj1`EURyM=D{4A<odr&a?6=n79p#qKZP!Qgc
z<L3^iU(T~p8<HgSy@Y2yBblNne`KaY)@HtWAMILMRT?Gi?6;ZyWcov`5Ss4FI<H4k
z|DnF9&HOEvShbv-OR7<%^9;4KnhWLATzLJGPIQOdV3l+Aks+mvUi<#q=l=ah`>^Y5
z!kJu39vBb#!&9mE)Ka|oJd$L}Y<a}VKp^5Za{No#)!3yNwt5!7)vHF|4&;hjgzk>|
z@N6i#PbF2uC6t*}j}Zfwhz=#&P;X};7M&@lM`z_}nM*XzS(H)o6+`yfMGnpRSJ?~C
z`>ej;FnNw}g?QNrY;GAUKA3d@m9cLzQDr?up$>ef>L1v*#o%SXR`H&XVsakoNL4X^
z`NsSYSX$FbF_WfY-_E0aR_|S0^$@ylM;cI)>dDh~N033>S$q-p!*j%DaG7)y>qT=g
zcK0Y$pNJFqGpESF@fTaZ$d<1RoQtBNr?B2908eGE)2;~~5G97NnC>F6(cTi~@cXOi
z$gJgbTuq+W6&%9vf?hssiwcEkc#0qO#8La!#XN7>15^xi!_LGMY`M`FQ6G}I^<NbX
zzj*=gT2fKoPvA$bKF)r;_>2LSijdv(2o8mTs7U&ZRz*c28{A8qmEF)aem2hUHKu@z
z655gxG35AYJ|?^8;`7oSWGep=vbAsdirA4DdBudapC}V0PEsOsn{U|lB}Y7D?qPCq
zX%g$?CZKNMcjOv;!_B}aB)?*>sM#tN7Y6&%sukKKx#ug?W+srKN)mSHigC%$PRONS
z$Bci4wC!mIlD9U|mkD0%i<2L|9XpJ@!Lj^ucr9cSfEux!u!rwY$>qDT+<6E^&D5n#
z%LwtW@GQ33>NB28wX-v2(X>GA8YwtiVbRoNnv?NGoTpL7>Wo^Umaeb0^<aN(T^B9%
zuFu4b##`LS=LPRNZ@@Htow-+uKb=mpp{>d*VLvE{>SPV@LDdfWecxfo`w0S%IvPt9
zggvL>9$^mKL^BF+qkF<1Y8llfI%8`CraX&jnznGd87VM&Wmw{=IGjt5r^fqxFwAW#
zX7$UW@}G_@HF7PDxS0nTYstd!)$PbL2q)u<%~;?P$n@l%6aO)ZYEx{)->&;{hv;l%
z+AZYY{an%1<w;*~1z)5NkcdS<VJN}$-x0p&+i~cN;~@F{80N(O;(Oe^Xn%DWYj?X$
zq1)!sBmF;u@3R$ly7tg=szmR{A=-SD5|pnGCr#_~q*^cl*8L=k{Gx?9;E^pfP1~9c
z-YJFMRfFhwnI9c2{*SiI(V{HRk8IxY`BZxMKlUQC5Ayrp7B5>CE%e7d(YjlSEEd=S
zPlCjHrJ9uR+gtq9ARGVg*z()Wc9?YQBu<}rM=izKg?q+4#lrAJQdsJR_hVB@J<Ex1
zih^-na+_$?`$ldyAPKV?M<6V=6DK<Nh&0yh6FsjTfxaI1nZ)r7Zf*IN^<Ffm$=SmB
z<M$1gESz0t?`D+J#~(_2uEXrC3qtIQ#lsh*uwzEm)I6z_)|k1$#>j$84;1)3kF4m}
z!NnM8_>iuu?IH2aV07M_51Xc6Os!QNe?1PP^@k>^1_WYiqz+|E>7!<2B2xUSX#Kyr
zh|auC7J|E@wPF>GH=98!5ye8DV;2H$$#7|If#xn(+_4)&FD5_4x|U#U68tH4P3hD%
z<~qV>exV)Uh?EQD;oIlptiG6>uD_<S%c|+wyp8C4?Fp40>VVE!B^;7j%RK&FXX7VD
zlaiu0q63=Qs8n+rXg-2}uV28+y$=dGIY~sipTPc#{@5!W$-C8s`f#fsP1Zjv^09bE
zac4?|p4?ke#>;xPx?dwJ2~DQLzLB*4v=7bNO`>0ks{HAfBrHfc3_qz+_$KR5i$e1R
zHo_GieNy<Y&i)Srj;5hkUwCJ}UyHE@BlxF|k(g!Kh|e2s;iI&V%j|W*eA|7bZxtnW
zJ`hYVljfk#uO4uoFLwLyBMPF9;Fjfdd<bx4*|}R-v3?W%yYE0j%C-2Dvy!@(1moPR
z4}8eBl|1pvMA)5hrFQ>5B$57!Ab5{wPRU@~6sDm2Yz<zc8ya1UcxQeO#aP*5${i_^
zcf5v#{sws3y_KDeOhQVNGK2{N(oZhs?-w6|!3=@Dx_&F`d(M`XtXIV9qq%hE_))A_
zEp$q}{4gTwx=5?fIVR=%1?w;euG<>8lz9hpmtBpAWtHf%j6q9#fN1~FO18yG*zG*J
zhs7Nin5^KgQ<0ockB!@@^V3yw{Ax~@f@2`l7|0XXR$!moHPO^7YV>U25VTti)E@Fz
z5x?RyaXPhHeE4Az^SjdkYqK;aoTpG@R7ty+Kc+J~SEAPBA=7uYhtbktx-a<C9*&IW
z>BoN1_~=|r%n^P+9<S(Oe;cOsDih-a0_n)$fha8YpyPE55bXAmrF}hx(tY#jb4(<R
zM+<&j%_n#_!wr)!m1DS{Ii22GgU!NOqU*f~1^LowSye*eIu7uxHbzW`COS?F_bcH}
zzv#1YzcyTo>78G+{sp(-i15v~YI4BsHv-4fuZ4~rj}Ud27V!L#RI;{jrlx7%;B)Ap
z*zrgdJuN*9|K=RJFhP}z76e0iqC8ei4HG;?arowcLF{(K79!`P*ylT4to+mhuTpiX
z?8r^36_~u13WikjN}1K$t|A8^hn}L;7j|#{@a|SkO26Ard*6h|*cZ?a{lMpJpV^th
zzAcj4BYy1yt$xMLJbv)=x?%i$>TSWBGn|coT#0Q8wrE>&hV_)MhxPX*P(9=h{bLfm
z*ZKj=T_QnA_G?JdXD6Drd(m3Yr(9uT2_gJ0saw_2eFrJ>3ymUUv~vTmwL<5wjV}mz
zOnO^9Fh70)hQG+AZ@shl&@REj5<5hDN{Tb@_%@RLx;%-Vf7PTxLwaE)9f@HIzEpPO
zE!A38Qtp~-=oq+{oyss{+q_>Q*HeUsYoBmH>Ze#IQ@GE4i^qoE+t`;jofN8_S-sC4
zOpSO1z18k?#4n1fW%ptRp0TlVne-=gD=j-Or@iK3CN>-F5-l`r#Q_y_%5$iuqMU=m
zJGO_(GEdr+6-KJ>zLVjJaD=_M596)<M7EBXDf{pvlIaT5di(A^I!A8Ay1!@WT$eff
zhoRy_ktd*`Bs@P?osoXdUc4e9pO=abLbAe+Zj2g4-wo8+f|cL-FiFA9sq_r<D?ai3
z1vBtR_ZXIW37$)-tGufE8H^6!rns0!s+;Kn>$fH}|693e!G33y76sD8v+<bz*_AR{
zhYH`sZIF}5!QCt${F&e)%s`its?C4gsMwC?|E)&TF%^0!k;}~wb~4#2F_Wye!LTu>
z(N(-gd~T8x<!@Vy2cItS4wG-Zw`3Y?e3}IZi$ORND9wlaTGHbIh5YpL`>ea@J0hGT
z;dtaXCAEGO-c>@cK;#Bf;Tf$Kd9c|vMzqfA7j6h=*`U`?vFgJ;==S^0zxAoY@9VYf
z<&QGvzdRPtEXTvyUIS0wW@E+ZbChma$HvwAup5I5=!&a5a=Y?``@lCOpLl^0FBek*
z8^RXOeZ#_<H)E8H6N^?+V=pNQMG|2o9U6{9?$Wg6#!FPUxnSH29pRiV^b?1T*F0d-
z%j91ckyl$fMwsYPZ{b*KSDY>`FyF$~nD|5WTRGp@6pQI+r}MIkW^7J%g8rB~oW0sY
z-xp58mHM;v)4&!JN7j%Yoxy3h%^12Q8+)35GI7~kLTfX(QoV*riPtIj(;Xf<`3N+}
zNf&+VO-Jr~DJn?$Ky#K(glIw~IZAY+<oI2dm6Snxp6*bM|BBCHe_8nA?;`J%ay&`v
zM!o2xkXgKj18Y)fn(*%YAQwU_{I^l(^cD)fAjfCU>?`oJg&j?n;4L3{3)U_c$P)Zp
zc{+BuuzV>^+*rXh4i|BWPm|dL2^&PM*^a+kyZEgNGxT@xqIspB@cQ=)mn=BKtx9Oa
z@L=pt`bTVnDf9|&vxpB>yy|x!Zsw9Ls#PhXw!tT<ZekUd$Pf;Fo5p+$uF~-uKm672
zt3BnyD|*#@nyjrOIc2}(w%3|*ujMyWswZX?;Y4{`8))JEYoY|LRu;ea1ndu`Vy~9a
zCotzcJ67N%loi0^?pwsx)nQ<ADkYp$X0`b@sLZ4Wag8!XitaaQl&>8rUrI;nK|`^+
z!W*vK@R-uid?J^BMlf@`%<f)vM&>met$Lwfl9lE~HgiHr(fby!-ZY;YCkTJ<-nSH=
zl}(=OBJt(bTUzXOjW$0_6fd_@re{4qI5*CW2bqV_kh@u!d-XL!`&#pet*0scqc)8l
zy9Zhcp{P;s7Tym#u=!~LR=GMBu3XJg7BYfsURaBJOSN&+o9Nk`MyMNaW#toI6+ZPn
zO@m6mA)==beO}g2JKJX#>qwL+8j^CC0*4AA;{s2t8Tgb6G>4M9zbVcx-h#VE7qNHn
zK<av#k3jYfaoS0E+wl}LMK$8LKVpIXX*5yP%$T7cl-O6wi6XMRohN?eJ{R3*wxQ$0
zUh3@Fh;x_h$jQQ=EWb+8{0V=Mqj3_eEmyFfLGB3I6@dIj?oeonL6z+^+}?PU4&_(S
zEptij6Zdzs#~6>!;qS>OMR2wR7UH3)7WyA7MuD&weUmj7&V3S5CAOu&^2yL!5P}n%
zF0$8U<Jo9|9c3;Rft#`u*oBvZ>tp#HR{PKmH+6*jMBlmWQOz@?8w=f;73!i96EjIF
z?+WP-ttV{{dzyYpolb1kf{cA7pRuHnm#(-%7PC)dXj&L^9xSlQ{8qE0N^8-7T_K7$
zFT{2kdosHej&}p9v_f0L&^h4)LJYra8R$9DJfXXmV73D;pMBxd_hq5p<4BRU^)r&s
zxWMf;#?wn7pEElqS@f^(4zik24pYbJ6ezHq8t+WSv`|eX4O-9D<s4}5QzJIYxsayK
zdqs<Xt-_$H0Dd6qEF=aPQ~HevcI&LbhEC}c>0N!ud?fcn;ei*fJ&2=0MenKV<-j7R
zhNTpGP~eJMtQU_G_JYcx*@Ah%3ZH`lX_xFZRyMVls|`Jh!RLn*eIF)|!K;&5_R=@P
zyvqs>b1X#dv!|o1co{})GY~r*9mSf2p0Rj_JH{mU!y&b)cz^K&U$@$Uk4}rCo6oIS
z=N7@M(k^tWhur1W6LJvHppCQ9Lt*z%2?}+lbY{zO@^unsx;Jj&-nqrJQO1Wpj`0^w
zCy2f83alYXdu&n8#gEs+`K}5(k@2Q9D3!J$)bBRS5%Mxd=7xf!C7*Q`|1O;0n<{Eq
z+5yEEznRSXI{IE_ime?2PcLi_>93X0-ZaBUa1O1ebUA(AbwSwY>xWTjK?{DUf2R)D
zeOSJH1Jn5&frR5WXzi~eN*U@+r(B)HeDO%SawD35R~B;kQOWqZYXG(O*`$>;Ka92Y
zuSbsgCnVM^g0}xf^c<{V6#?%d8zHzcEncxSy<)T__7mnJ%h|5*foK+OAqCfN`nOaM
zmJRBnadTE9{7o=-t;j-$^8*x}%Ru$;TY$%D<~kz+GFJC6*Lfg*b`7DDfm^U}=gLCY
z7l)y9dMO3>+-DDjd}3pC8(Wq!hbvt^Lse4BRPalMdYkO&`WAm0bGs8tpC>^vdNJe#
z_SeXu)v(gI2`{&2Z0m-js5+d1upWjfn`)38G>@Jisbz%^mebMGKQLjiD?BC%yt%@C
z@QAd;+>Qpmer7W=z4BPb8EI_~gIIRAFcf+khPb3H@DY@!^P1!_cx1R9B}=Zby+ee%
zNxnB?=3ZfccMZ^<mOPPy)?_2*vEWdByo+|d8&44fx5Fbh5oT+vNMmdaQ#F?;+N}DS
zm7XhQiVEp8wZG80wf@N0B^To98E48o-pS=e20W!O9?MGK!BW?qzPasVDxR(wIXVZE
z^KO&fy!)g(eizO1uf{Nm9+qn`7!NJe3Kv^9hziZ($?uqy_JDbkOv@{hxW{DXIB6PX
z?OZJStPk<1k8NBv@&;R$<cqn9Vv-pqBYtPRn6$LQk)~xujrk$a*mn~(%Z<oPW)Ws8
zr|=yEv$*OoU;HP<p<w2$wUT9sHTE}Pqt_NPEesTms&vO3b$8xnEl2UY!};e^OK7TE
zICIWC!V+iigYgl48ob^W4;S7=!p|OPmQ1E?^GxaCyaU)?Sx%BGvk<<`4$mCS+2T{V
z%+72BWq1-)@-uOE={Gh}_&wGq-e9-(*t399z+BaCbXTgQ{rExAtY13x*mVKahrPh>
z4<V>t^^`XE6L!rPHPAXqqR1gGu`v1OCyYBOq3zU~jc%)K=Dz+d$&5B4@k%9e|09nP
zbbUBx2wX~;G8dlIr6%xcVo^KzGEW>OTTn6W3fyYL1kQd2^v6#VcU;|xtV}QQ7?Z);
zo2UDuGEEK7^OK<(Sxm~e878|%W9*_2ylA=&{yvT3AD+qzPMXaa|Js6Ptf<889gAsT
z&<h^D#E6vI_Tm1kGG3kOh;(OrBzOsoocLmr-EL1)ruC(nGo@*32Ql%rSY$u=MP040
zp<{cHTm#lq&Gg$S?MkIDZ^hVa{uc(z#B_OM92>S|Ep7TzPmirbsOn-E2AXQqL7xl!
z-ep5PD=);Y>wSx+JQMar32SlaTpU(CY~mZ`*YNpLTiD62K1BlXLSULTU|Gu&RJ~tG
zZJ!*Ozd{UUkF$f?VF~fH{Z}C`mq0g8-=otqV`<RqFd7%zA)b<NL&JoelusW;(S^_Q
zka%DZdqX#_^-v0<FG$n+5e@<$B%O@D3#`1^v&i+MGr!mRmER0KF3ik-vT0-bX&0}U
z&j%QMMeFTMNZ4|caHwWSR|nw7vkt1CHj56OWY8I+&yw0#^U^EAH^aXI3#R*0bFBpi
zO$!#S=~|3WYcE4NW}o0@$%M-MaBNwzj*=(pW2>i(kXZ^vz^)=Xn)8nT7~X={XLF(A
zdP3OApGJ>kINkE?qT=Z*(RITQ548xgDVMqBT~(N^2t;}9HR^bm$hqk`WNt`D<cg<4
z!*m+cQWNesLp(9;;z;pchfsbo<uiL(@mlmkw2jsZ_Y~v(uQ9iGG_XRIZccc`N=nZ2
zWs9W6%RGd60H2GwPki~lx&Hj&Z(Hp7xsb~5@50s-F7)MrD)&3l$@Wir%$$A>)jnP2
zCUQxO#DKF~u<T3#`Zu29SLRNk|N5%%TH}i}vn`oTwXcNSW??4xb~0PnPn};%P>1Z~
zQ~2FcDRw&@iTFh}cv_jobQRm_$FcF`SzyopmItDBCeR}6eZGb#qULNX>$9we6lNFj
z#jlF^l&zL{5THg0TP%p_cQA!fw?t>}UlaOgDeT9}D4KG18p^*^5^Jr2=cgVNPdA0G
zw3x4!AB{-s4DNe+6aDw#H8vP6LagO>dLVRYI}c?dQ_hfn+Y644Mh%=?X^zUCY?iLJ
zPT*}<(ZzG|;)Qed3kPZ2(QD5-fvqNq!KsqkO8+|fmWT?HRy#~^^I?M}+^H(rmM1OU
zgaAo-bUe^Q*`)RSe!SqJ=`EvgFBf51_9p(DWfVNjY-01y*kj@EWU!D_Jh?Lwfu`v=
zVc|tv2A#%!muR~Fct4D$<<M{Ww+LyC;Y;7##M?qy?J?b2WMI4<zqf~r?i5X-=MRri
z({d+hB}o+x88{XL7CXSFL~x78%%XEvYfxq5ht+nztTfa}6#n5moBO1dG-sY*8pBVM
z!l#!s{lYUcTXc>3Hbsk~3rpDeH6uwa2=G6XKx5+jXs=)RkhVz=Wf~6ewaowcVfq<)
zYO-yHbK(sID-0r=*PHQLSD&U1T!H6_)fn3^0vmK^Vo*&Crj8*p5jcgtQ<@-JKe(uG
z)%U_{LYK@&IvOf6Yv|FhHLP{XWBhR0Ewb{uh9K_E|2(O{*#dp0VfzPO=Z`b7@}I&R
zx%<(zzzyv_gOIsKjihzmm?Mux{(%Hhp|CgH|DT+|?YO}W4mi=nokQU@b`rI!&4c9;
zf6nJ*i;oqTLg8tjqPPi0qQn;|^s0{yF8#6M_8O`f6`n+O3EsF8DDcTOTQM)vRp|Wp
z(eAX>pj*Ow)<NzfFN#jUu5Z3HtW=Xryiev{GQVh<P6Q2V+RoHH&8cel1m>~-0W?+i
zp}({*?m3#^qNXiZG&MtW-F+_Wa1teF1onWQ4JCCR<!h8Q2|Y7d$iZ|igMsyAm!v{v
z`)6Zjoel-sP@(G9d{}<5ftJlBY~6UD&RWLMxX$tL>H1J;>d?gJ#NWh3k9AN_uO@As
z8KhaY8pe5^T=`iWJ`9S+5s#Oo(x^!<nxs(YVvVZY6{u-UqJ_BwD0W>4?#+!6d|11<
zTU0du`@5XVkF~>2wpU<D<+80+seG8sP{;?D@r#x|WI>ZjR<;&#Md74i8iq9miG^-`
zE@8t?Wk@|pr;NsXP~TOI5GQve=A405i7r|+UW)UU{K3zqLN@8pYV>g(%QX~6vXcu(
zKvFp87-tB({BL$5GbtUaIr2=LZ|KhDTem~=tk4;?VU+r34b05n(6~K*u)O!T@Sl(?
zf3vs%)iK7@rV@xhsSz+AcO8}~_py5N1|%G)qsk?7NwQ}wZAz<U{UX~)NAM5?O5Epe
zRtsUK8^@<APr@cUTY9n}h1TxL!`#Y)q;*S<7d~*oA2&@>O5H+dYet}{!;QSg`an9h
zijDYE2{VPed~aV@>MuVJe-pPO`Fl7e@9ayVd%rMC%RtPMbEWTxc4NlEQ)DcCn|pj0
z{8qo4;g|Ct^!08CGxAZSk{ky;?VGINgdJVDVFerAXEZ?I)l}Z#Vtb)arM{{Xlhmxq
z`>k+x-!dNCZECRgr7$CW=|dw9_`qsJ3g&<2xU#ttwRuB@&teWGf6hVK@na~S_ejXK
zmU5#RZZv3UG`?P}<nzp4veEsk(0;lX%aS9Ia=wEnbXdS;#U5&Y=}2}OYxv?{mM{^r
zwC^Izd47csJY<)n>gObW*Ep9w^$NzVnWqX^b`=fiEMnW{dZD*M3LnyI@loK~e*KdR
zm-lPP`dK<o?Ji^T9z6n!dLz;0ASCcvH0YirKAkd0o=yW|{Zuj8%@d=J>=7vx|G~4c
zK}Gp}IMTPJvr*58xjqO)o0Yrp{`A3x2W1!-e4PA3+S#51YoXIR3wPHLtvptO*K)t9
zt^Fn*tJSlkx5x14E5bimw7~DfCf>ND26slPP(n=+Hd!7;Ye{V3ggbv(VO%4&E(+sI
zqzdp~XEZBX<1ftNS3<*LujtYkfh}D<6m>H-koirH#;x>(tW*)6@1BC^laqvgtQ&36
z5&VPu`)D7jN#O-geff7M;rF6YiL_1bh@1WsZ1+_P8G8?T4N<)MrVib7ilh#`LO!K2
znRf^Rl#*Xwl&oV6lYysc*H=eobMLKq-N>y-_$}-Qt0R!>HVDs6tnfqg1wOZhv*)j^
zFm9|Cyquf`E_JlHckyOYKc9op+;&zwwG^N4Z$R8dWo%eJo9e&m@Ce~vN2Q+Bo)O5U
zUNO3;U`aK5;&`{+7Hm4MPAk`4z`1#QdF~U=Jkk%~fu{q-J(I=DgnBlj-(>ou(#uLx
zeR*QN80K@W(Z16g;hnD#r?(B;6&K*miBG&ESyl8q^&W!!1inSAuDC_W@+o%OQsw(K
zOn$*J91`w2$A|gRq-ZJapBqK|hr<iD*~zvr_Ng2tr|rX#IXPG*^%ZHq@@Z64JdZf7
zKwHDzsW-)#JH0DqYa0?F)jkkkb?nhw?2q0DZ)l*wb#mXP#239ULg}Yt<S6e--Z4vf
zv&~AVE4rg<k|B>bs^@D5=3#QNt5(9dWz?gV!$x<M;MPxfvYSBIr&EV|qyNb1Y9YS$
zBvQS$;J0Ph@v%1>YwqW;c|uQ4?xX>!|9yfbFP|cKa|8yTp2Kxj-7umdgH=u~rHuF~
z{8I7+h&JZ4S;1K}zPb*V&$(ij*Cb{!Q|K1mdCt8Jq%nQ$6w%}RE~4}@dwM$e1l`zR
zOH-e}=3SQ88FUMAT<a&jd}WNcQ&aI`W;`D;x1N6#<}ZzRKEpDpm!DiIhL2Vvb@X|I
zVL6!s$0Y+5p5I05R&^tCbP!oQ;c)vm3YkWe;51*DQUA-}#-R<QG4KS7&wR;8M73be
z@UQ&R!$2XM=1vP1H(>IH0$i4o=0>Ax__}|>d?i$sKE}5qH*Prld-c0`vX#(1h>D^$
z$JA(SVLFTt<zV1DIr=nkJWcL-kE!KeFzS~pR=TUlbbH)ce&a<tlr{oYdxSZv;xNp(
z-2l^o0F)mOWaY;mapg-7DJR;B2P)QLY{y#q;1daz+PU;`eH5fC7PHe&hEYyiDq4~S
zN5D1V`@6Oh$Fl=5L-6Z#)g7Y^N$>D)iJ0$65_+1F5zH>p2i>;aOz)lpS(LkD6~9Na
zxqIpTPj#C7vtQwPkp`VUV#BTvX@#D`cbuu*k5&GIwGW4$r=79h@Sf2hhGthW_|Pki
zJ|%-?8k?|Xx(12r3~}VdA=;Buf?Xepe*H4ynt5G(_hxA%uK0)*l8^aBwRCQ;HvwmE
z+2BW`7o|^-;8n%l1){@8xYv_sqPX2x@Xy>1IWotw>nfv6=V@$ow?7U&QJ@9m<hiQf
zJCxrRv7r6Gxy1?zX!bK>UCm!aHlrLw``%te!wY3LrVk@!n`O`xp0hDa+R(bsnQFoU
zNpZ?#me_wHF4)PE%%hzYIW$!KyU!d<R(I3-+mS8`uv$*Hl{(m;C6m~RC&Q5N-JbRi
zlffHZN9O09h11HTFyqh?N^8r&$`i}!)YC#%l`sSbA5xIB^CufI;j~t&@&?otd9WXw
z9Y{-a4EEma;M)$2rGhi*XfbmX&)GB-n`W$}`WRtWa;KVpy$WUd|AylA?PxN(@qqpL
zI2v17ETVs1;U$(0RJ6$k>VvKLvaq?-EOhwzrB|Y$7d7n7#5E|~ah;7G(NK6lDFLE|
z{-Ukr{j@!z<CsEGChWD65PEYRJNU*CJ9HlKkV+}-Gq?AOO~QLcDXX9HcZq+PSw<u@
zB9&m!SqBCGr(`j_iD~S;E&go1nF?+GV^bF!^GkcLV86ivw(f}r*;Fn@c2YBZ{UiCS
z`$4oO#gkvJ7(x5Ccq3BiJZqGJW(aq!H|dR#ocxLpemRuIniz{EFAk<#yJYFv^LTb5
z?IPP08^vX{-GwvC8^}kL;n1D8cw)Jc&S!h#+_G=<Jkx`ZSmA@jAL_K~$YAXl-4IB}
z&KA0Ni)ik%P?RsZ%)4e>WE-bfkyMZd9xRsNV%=`M_#Fg=c6YY!%q{GApi0>X{IKl$
zFX0_}6LCrjY;cJ^p1sNtynENMz)l7qyneB>yNd+&>LpsTq?zTOHW3(9PSkt+9z~8l
zE%vb+uAP+l4d*5vENuJNifBDE%nuAF^WrYL`cnZzL(Z~OkFR0gr%C9aA-oF@yrLT)
z24iM^6Q<QTW0+bIw4xr-`?dsBNhHBKC57K}yvLp>bhC%H3gW4&B;j5dfZNJvsl08Q
z$oAkJp1;PB>u+x%^}e;D=fy!VXt)gzElKE_%*3VMV`zQ1g=!7@qFr9hKlUSvygC|R
z7Qdz3ccygYauiZ;9Dzqv9$o4Rqm%oq@h+u8<US`*@L$|S>c18iQ~d)*F78lSG6)H2
zCNx_j492T(iJ`a$&*$%gc$<v2-nuwe^LnvxZo5FWv2XZ~!|SO)aH=mj`T_+HeA%Bk
z2Yj$R&E(tU1YbcDGg)yCB?>ysQacgh{T?#^L2eYe?u}?iqa3?<`#R}6n8Ld06B_4v
zqWM2<=#8_%g6-vGUl)&6%X7%f;R`zyl1wTBmv=SXnEyjdQmPg@k!`nWo0l9m_|zex
zK8BYWI^b}dJ6<&`BG)ZW{NW)Xe-t!@eA>$kE#<}{cHIyt7OP>}vqoCF*qRI%Ey5$)
zLauM}jcNF&pswAX&l-Ij>S19RWt%D*@%j?1OctY`W;q*m^%yoM?B`PdBI!l4q_*wd
z7D}8pnwm60hf8HyRC$h&=bR_j^u5F5=8qTYcL-dTV?svhN-Fdsuc7~npGaILi%_R+
zD7+DdZND~)&j>EB$P8@^dR>C3c|gI`M|k>X9^AX;60LQhOHY5YzR`2(&BzCE<d0D@
zvqJRT_7c(9E7&sF8XITS)4>8eI2j(K{IqMR*x|(1Z#5Ku=h2X;e8(q;enf8mK(6yi
znOfH`E|faDfXTYAMK>EjiN43N{o7mi>t(V?(Qzn-G}yEFoX@O}=?rRH??zHf_F&*a
zNp|A-M(E^ErE#++AjhPFPl=sO)h++nmf!(+o1BSx;%ZU&r626b&pZrWeFb#Afz*_<
zapzSN+5Pol7lJlp&jEppVX_2mU+S5BU=_N4ts~?1G77%b!n(&^MvI&@d+hj~mHyMj
ze+|L>_xP<)y&6P4iWT_AV)>nrW2hW9j#|9_iBzudrtpbD;`(L6E?E?SRK>+)dsv;1
zoN<*KNh<IOYyE)(3*n!+hg+(ul4aBc_#K-|fA1>N2crw(;J|Rv#*@!+Df9v!vz53p
z?EpNs?1Z0R4ZMsJxSUkKqGv;;w7Zj!gWsFX&iiKYbLoD3pRiwjR%p)mEK<QIBN^?j
zQ_f)Yp;4sZ@R0m78<Cw^L2=$Og>6H(Lgrw#s6{maYqp28Nx6O4&i<7MELCIQRt9mk
zb++P$G+|G*=??A6lt7)gJas*eMYo8H*Ice)8d}bfmz{?Is|nms^QQO%9ieL`Qrh=5
z60p#~Rvdr47+GeEsWfB@v{W1ssuzK(Z8xy+-M}I`?noi-kNCgJ1k|U12b}QXX(_@n
z^}kRQT$)2}rBCRYj&#uinSJm|RwE~qM1f1*LZ|KuoQlTJyyT!V4F0`f?KP1+A@?Dy
zxR_RKa-hqfRcQXGF_<V0!qx{XX|vof{?~67N*8bD%eFkC)E#1c`<n&bM~(tx%tfr#
zyq`s#y2;;9$s)^&p|secm5s5fWQ`KS|6x=u_iJ;7m%3C@(5jUrw%LLAgRHT6iXStX
z`j~z2%AlEEim;e*9Zjnr;<wQ4^|u<%a&j!FBv1vfJzbFKm&jB1bGD~rF1fA>$7Q1f
ztaQ^)d>_Lhx;cfermw(}`?Iik%zV*-6QyK))s;)!@m$g;)QkMaw1`Jtf2vjeNAU0+
z?qNgzZlvQYwv$8tI@VLLi;O(K@bQ1j5vHuh-MyWK?tKSSsvS;-2TR~2J-8@3{T|aW
zo<TAZ7clu&A@6_jtiWU6fWCK+i80t+^w+3~swb`#?uJ5!u_c91doK9m1-EeB?AMqx
zI-5NzoxuiLz2}z}oZu5HqR~BYFljiuVMj|9|2*RgcHhon{wE7ie6T-VkI1LJ8P}k3
z-%xzcGZ@;%HKI$Km*R45H;w$NiO?ZesPnqO0ezUyD*HxZ<%{)9@n{0A>^(%=lt+=_
z)nT|jd>-XD3Xa^S5|T@F!1cIl!Rczs-@Cp*e}4`7&>*e-tGSX7AOA|!JKGp1q$^P<
z*(KIqk^o_`#f8v1YC}yZ()*`q>ta`&n(N2M)h^@(rYdaW?|pQ|SP27tv$3VemxeUS
zu)_Lq%+%UXq4tmY+ocn^htRW?noy6=^F%!V?tfZsuL`ke&vIIy5F$7#ui?L+)!0>b
zgYxeiV_3Q#)6U#Okv)cdV)Wo5n@w3vBI^?UlAkWj<n4*IpW`#Th>bW<g3XFmH0_-P
zB+q?kg@(qwDYZ`ER)mp-z-m=$m%yl1g_L?(2Q|Cp+4(FpYG|58hZpYRgWg$io7Z2o
zOxzD(>eizO`nL}%5k;bNwpo0Bek8I+_9^O&_oXoT_qh4&3eB7n%H``aa8Wk~ZR*0e
zW@H??Hp{~9Ng;-1EMTVUwrt`(S?xV>0k~THiYJ#%qTf4*Y41?^Pv|h|aHWiB<PDxf
zqa8<~?93#rjeJ1mM?AFbC)abG8`%Pj=qQ;zIs~Qs9io|qc9aoQB0e@vj7Z%(IOT21
zT-5H8X5SWgQX=jqU4*)&BlYfHPVR|W*naL5nb*tUb5k>O$`|%T(OG=UlWWi)DZ<dQ
zQvNObC3CWJCl8A*b}M}g&py$Q5?&Wz%w-3`H@%73M#+;-*f^g4eFBxd4TO0~t@umB
zPQlA~4)ezZ;CH<YuV)6dKWQj)71*PljixNrO&4c(En@k9(+idU&7={j%H$FoL*=#y
z`LgaS0-M`{jNgqHuZS5+$13)7IVESoL${i8f+T?N=dp0qM?7~8$0K15rYkcDCB84P
z|8^sJ7{$_*x_xxsa0N|Gzr!~4+eJ$|Q?O^Hm@V+zLE=+K=y*i}?YL-+=oR0zc07BF
z+xE@qH7P<+m@<1|CCtwM{bTvT6A-sta0*$;Xm406O`T)CMGpf$;PQj%G;Yx}d@f^r
z`=m|K3(G}mv<a^p+=CsVUIHt2E_*k@hVSnB!&4{!q$k>TFiPEuH~b|$Z`soC)sIE8
zE%8k8jD={y_9U1teMS|3v)H|~N>Sy?bPW3}@X}lbmc_Nb_^=}w{Y4*OV}B2=FVbkX
zvV^wDKUaeH2Xv-c3p~I}yj|fgJ|zpej<`fJ`8u7peu<?6ccVcMJlMlYQsR)<`&#|I
zg#PA_E@TPk;$9iSL3{c=UqP{$P+thA&A#Fj54^EX@dZyCT`rca?Se)0I=XrA0-mh*
zLy_ftN<LS{629*yoEyMLocT%%ItSD8_D)uyKZc5Z1&VoW5;RY~X6;Jx^lV0&z%Vu=
z1&?61%-fF5cUs8?c?g`<92N1{Deq{O`W4zWcP1ixJ|gx^BG30#Md5)uybrhyo$dW$
zB(9>nf1*)x*9O-zzcVBE7%2IlX9cd!$WK!)ob>!D`@7DQejGBOQi~kX^q>i3b+CjU
zmpfuof5DxZ8US@R9OKS(VH4fQfN5XgIU<$xvb?GL<3LmxHuCEeykMbOf$e=e#gl(d
z5IvK4Kwh!7++o8@QPH(Iv`;1yE1UGNFgY0^^WLNMmlnMivigHP?_=Ng8#GS36h-f6
zAfQJFY{+7kKQ<DpBZLmIbUCjaZc5Le_AhFjla4nNCPJZ4IqtmA<JJ3BDMdqg=afxC
zshle1<|(mE<>55fb2Z#un)u!;VWhmKiPF;+pzen*S$}MxrE<fHq=Sai53K<-Sz^29
zsF(rp(bz@$JN1YS^umI%Hgu_=UHoYMJ^n=d7oMn`qe~`NvAVj7@_K9NLuxKI{V;~x
zO+9!Y9Sgt4XDnpo6&@t@nf^w(kiDrM#>cF}qd7msuQ#MXZEqgdAM(JvY#;IO@nxj4
zas!QuF~^|dBK}`XE);s5g)V9sSM0Q>wqJvgHK~;zX7+>ba<{^QrxHceGZL{hLXu1!
z<8e7Q5gyOInWnTUmu|C1xcEIy^}YgjQ?}^K&RVu@bRX@8zq?57-b7}zA)b<Z6-8Iu
zZn5qishWnmH3%rZgY&%-MaIkj!n`1yxjzx%@XZh?yl>;7^R9}WW*@;ejK$C~MpSk$
zfR2S`Q1{Q5qBB<|wSUtPIyUSC4D_y`&Ci{-w<crQh(x-3L`*ks?F7nfxYzrWyjblc
zEce&5DF-gnBlBf6=fF_335(*cjJ~kG9D=biE(r2jOu7bd`IfQ}`cg55sW?od`ETFz
zF_XhY%V)YEG)zlin}tK(Js+9gk10@BxJx8fv3K*5M0z{iAQchF1LT7F$_Y8*V7nMH
zk!fXZnQ~-vUk#x<qIq6V6#DmD48Q2zv~|G_C>k$^;^J70>sQQ5+ftxn+=pVbw(!Y;
z2l-h0k7$2x!e2WU(Yg;eIF7eqj$JfrK3^8*X|H(q#RaJH`O8I<8`1k%;N;|ug7wFC
zo-|ub^?QA2eb_{F3A6sahCRq$-^~(hTTt~Z5bch!qPT^-DN%NVXkz6|d>U~LUV3h<
z#X$~HgLcx}v>vz$^Wz|oXuK(FMC7u|^dzSanFBA0_=COt)sRoHw$nuVnGJM%JYcfb
zoj+1?5Pyl5hI!s9(SN&7q2Q`H?<qM&D{|LUPm{o!`C1Eo-bo`}a*$SCh)wFNG5lvD
zt?ZtJ&AW|R$<KGtJ#H>=(mB`K<%jcI<Vm+;5Q0^<;g`TB%>5F?G7l|6^gv$(q~Cz<
zxrub$&KkB8KJsI^+n7{_FNPFYVdWV`q$@0Ct#cw!<WNa=FQ16q?F8q7Q-V1C)f-`7
zxE?Z_-V~~d{y^Dc52c3(BCM{G4YlpT;we%{FG#_qDMCIyX({@|CBt&!5Li1oBV<)G
z%koa7PZr;K>fIkAA1}cvGwd?yRAu3E`v)G~Xf5!-Gq5LpAoMTA@-vOL0vkuaFf(T~
z4H|z#$j=8-sL%;2E2_eT?l|^M$WDGbc^?6bhQlDQmNbL@GV@`>_>>1jp``SYq@^SH
z=L>G^L(mXjrZtBoF1yfxFXn9fsuj?G(E>Ny&-f-}+Vdnw<4UC93i{HDfL9jqJ?M)Q
z6H`Q66I^KHo_*}qiMN#Xeh6=?a}!(#MszM<Eym4HVDY*u(R?eLJjV5~8xxmM&ZSuT
zFW8qX)D@`jq}O<o(SQ?SUhwbpl;wYr)Ly9hoQ1t{AW_kG-sDh3E^CF1%?6HA#})L5
zg`m-#qvyT|AG8J)vCC%2KU@wbv4ZTckHf_C*J;5_Bl20{EB@ZWVEUg(?A<pS#t{J|
z9yCf^F~1xqs-(4F2)p{EY9|buDR8>9+IZynZ76Vbz(DsT#LX)e&u*ShX1n9GUOv?0
zk9U2dA)O}$Pvsf9@3sY*lBqmnkQC*;aAj^ANZ^fn(%<-U>Z~=O*g?WQ*l`8F_oh!#
zdEqT+%U8l!QI>8z+l>*<Q8*p!f-zBxpw)hb;vzWV%M=!s7mL+j48#lXiRs1kV(NQ$
zvgp<Vfw4Z?3bv0<(dyC*{9Md(EHsS8%JZJEmJZ{Q{g**9E|BeSF2yy&Lb|ef5%x;R
z(U~Ylrd4|(nV?SIk+qZ%Ec7Jw^aKXn9@4FP&v@BA;jX@y%!D~{&xUX)7`T!3<JX9O
zG(bGcM4s(Aqd_5RZv6A;y>QByPZ!ob<yP(Q(D2BTRE<_p%4K)F-Sk`J;j%$=Hrb80
ztdSFo=57}<JEqKS>0S7JyotX<J_wweFBr1BkmSpnweB{HF?HHi{-*T|&YfuH8?UKC
z#ygU4&I_RMxmM)(-hh7@E34i0H<Ows-{7?gJ)+Vhu5j%Wfb_y*HsoX*ls5TNz=k>S
zpCJL&)-MS9I|jXT-Dum?UxiC$YS`nqp*VN1i;Y(dr3fo+!QZ+Hcbu+r`{~ZiY_K2C
z6Sx^J(fgp>5Y4SlMgm^`SiK`raG`dK*15c;nYBiET+z<9b!$LM<1Ve=^`<~B*^Xoa
zXFw(|2P5@M*_`r6Ecv#*&`BACpMDL<R=y+bxvyiTP7G>}+-0KjuQ+ugj{UokLMDIW
zXnlHyR_dsQ^hKE0TgdJeT)R(UzV#?GR{6*K$<(tw-ke484DL0)0o}en_%}Y6YyVq@
z)gu#lz-AvZ+8ze4q7J5cZwymTc|%o|{k44zUQn5k+Zy${Uhu4$3O%}5<R<UJx<|uV
z?dW7Jm%5MaKzu#7E$Ty)THCO}J^-Kp)iAsM<@CHJk;ctW5uaIYM}_6W9pSf-*Z*`E
z_eN$x6dx}5z=|NhRvE9KxsYu0a0Cr%MBZdI8mxSaXY&ohy)=&UOb#KSqLIbf3w`hv
z1@J#1uv2EY@aw9^G-p>5dCP2vfq4xctT@Xb&uA5Q{W*h$U#r;MSNSx1vndaWmBk`!
ze@q;Gh-+mg;$P4R)*WuDwaYr6tIMuJLVPSHoqUag*MlLy#ECx_GO^)nAHX{%mHIpw
ziP5Iom?CtXRvX0Qke(5@>6oZ_dy`br>a-y^ndFX~pnXvDSi?fp421n{6>3KQVEv5Z
zDIj?(Ql0$^$9WCVei!%(pNj88s-ll*VcruQdsoTy$7e#Owpy6wedhWKsoX$KgJw5O
zAhQO+Y2Dw0Jm&mI2Lw*axb+j*M90m{R#J<$-HW38|DN*zhj^0J58*0CJ6K8K8nQ|E
zM%968VnW`3-GUX6Tls|i54mE+y>~Rxs0J5~z5^4!HN8dttaRQ}wCS4A_M)L|nz<rs
z8iWky*NGJGVvJW2ouY?RBPc}3V+Qy3#RbVHRN@dJo^bFmRX*`Sd&^;rEA*kFrcX4$
zS62JnoJ~kFbR+(J5ry>s$*iW^^DG%1YEDjOyMDb#(!31zJFQJ*J|>AOZoBcIZ?D)6
zLt#(u+(NtFIw0t15szM4P$0p=5jwI`ESs+)<Xb<WS;*rUeKdf?=`oP!Nm`hGgXTM$
zA*bOacRF8+yrQdcEcS)YudQPD3kR`ojXQVx8Ac5q5#sRQFQI7Fh|5*p*rz`RhAU^V
zf<*;PalJ41JK)Sm4!FemKruFYI%3nI>*(<th8!0Y_?;++L&H14g*k`D7JHMtqHK}0
zPA2_wP$U<*FX)VN;jX-btuprJ6GvWWnnMPOWVXFvtFA1dcdxI~C_fYAUAsahZfp4C
z`H$Ht-v@kYZHbVh4o8+xUoxK2f_#ZY5(-avT~>_s9#vxh5AP^w`wJ+~_NDCCW;k6L
ziF-3`xW@|x7$+>i*PbD8(Y%ABrVaFHh&#(^_=5a0q5Emjk1aR4&Ngli!Ww01ZSPI;
z)X(iO>F>YKCJVfgKQ8leXO;uHGbf?<ZV5_aUof@Gzkr6|?-?F0mXhs6{;F_p_mPlr
zxfr>@BJ6pmN?mL>lUjcjBlZ|jt$RO&s14BGwls}w?>f*XlO~c5_9U^SH(R2jNTvs$
zh<oQPfcAbBa=4y^?)j%_vgRxbco+a<2@^g`PT(%4J%Gs}N5uY4$HYrvSj!#JnPtdU
z>n7vv?P6$lsc^>}XZV#Hp(s8P0Rw)DlRclJ)zP+aT#h{7nG%eH!>q~ZQz+H8U55Uf
zQd~c|3HNjLK+<*8)8&UWSvyQkjAq~beet^cJ^6cTlFzT5Y=wg%y~{5akM5d@;v6Sb
z$EosaJqhio^H16Kl|t5ekrlrt7V>=;w2|@lD%|Ei7iH)6p^}-w(3w7tgoq0+D5kN>
zh4FOOHU#pQBt_-pA~99Lmd6$T;Rl9p!uErOnA`rB#-ETZ@>l>-{bqAa-zdhuim!b8
z>e>A2_bb@3IE4By%)x<kC$LUBOtkCZbDUi$^zp{jv2PoE(AaMi9(0$J&ndxaG5RlC
z6ez`Z?#yNr^?!)8lEUzJ{t+&pP%G>RMj*UBkgt5ciDKk@Fk?(C9LDzGe7}!eUrrGo
z8~YV4S(wfCB_(p3pRv%|w*`Y2yHKq1Sd^_BAX=vsK^N|4v#)dfNdB&H=1i;OBmRv=
z-Ut_5`#ur-Bm(L7SAF&=r+_^Xe!h1QNb!#&8#i}6Gg`NcJ+!fB3%5l=XIlp%#~H9~
zuY?}ej}#uUB}MqVreJ(Op$pIIu&dk|GdC9rd6{x%Ru(Pp*}k0VtBP5sR|c++6Sy(4
zj`V4DHohOR<jq%9saYWwpI-#y-u;v4+Wk%BG@7BQW*{QU12A656SQBE)HW1V^0;MF
zFjk>ZJhRS&$`*|gI^xZgympho_L0QS$va6WaRw}ULba3~RB`CmHtzUp1}^FyrN7nk
zXgw|>^*4d+Lh3c@PPxx^hI3wFu$~%zsM6Op^<1sx3l|77kp7>d^A5}LeZzPfG|lXU
zgpf_B_deeWWh5(wjD)PO>=7YpX(?JNrG%tXNmB296(zKVRHBfKk{Lz#-M{}k9665X
zdAjfGI?vB%We*Db@`5FGJ|fX-Y+(~i(|P!ov#>B0XY07rsBhFqPR$z9TfLLw75<{I
zFq#c*@@9W*?y|;R2JEF^NRH_B8vL~vh89V%+24gG1f8PPV=*uqCAj%(;*suD!bePS
zgI0gR$-O=nGrOLGywe|+QYE_7SADZzP8ubd=n#yICLP|jA)l6p>hps?BU#$5zI4!8
z54~UaLAf}$Z@VDwQc@qt0&mjhy;G_2v2aWmT*gq3J2XZ4Af<hsh^7(hWc99r#+-SZ
zJ*5k%{{3W>b9;FAXrtFBu2cAaM~c0B0#d>2{B|S^zZI&WIQ<iR^J66)OKW6~4<;Z(
z^$lWlroioz8tlfZBThAedRX2?u0jT1K41|tj`ydZ&QZ*NzZW?thqI%juffZ%5xScD
zNcY1sKH;Pk>u$_~>9;d%sc>VNEl-eT{;J3CQ*%iDa|qnqq6L>N9qpPgai{JSa*`TY
zdw?!2CO1+(GKCdhY+?bLQ?eb^*5TPhA0%7kOZKGCz-o&WGIiZS?mfEX)E2#A^DR?w
zs6li9i~g`vg(5d{Yy$h#EOJX`=G3vdo~I=!lE(BE1lOl*eytmSyT^u)JD38expDNX
zz=6{BK?XH8C<*T)Ejm)nu1@a3JDwX$?8N!mZ4XI?_Z9uRpjk+aZ(@ViCSt;lZ!$ZT
z$7HQPfRD}hz*e0ywoJD_{wQ@9_s#Cog*MwMV2BF+oaset23N5IW<+jwhOmwpM=NKH
zku^A3^8?nW5%wSh4tH!}-JwjYgK~Jt>eEbXyelPEjHmCPqmf({%_1F6P>t}z7`3}V
z$#n%ZJf)P@xeW2k46y#~al!ovkPIsRfOkhb(_ZC9jQ*@D-8SX~OW4o}H4Y)Lna;>#
za2`q{zA?VAKYeNy4Bzahw9D@`hK`6p{Kfzl`ZSA;8yX=w>f9TOgPetT)r+?u`j7N%
zo9J>;KJ^?ZSd1m<GJ`$agwrZp@BkI0d*<B1kG}PyV_?Je`*!E|w#;N&dvd5aGn6H@
zSks$VKG@vj0q-?4je42R;&T?{qUWQh)Z<|cEjVb7J&P0tW5^fpQm3M1)>Ym*d>E`c
zl1VZmkTw|TqHbyo>BM!CHjN0T{#ILPZuwk>wHGDMx;gl;_Y7Xmu%_#3nYeUq091r$
z+oePvxt@ae-=>k{ml(-MY&*_pj%(!}E1!Ut8<EqkM!NT$<4dL=GW#izwB`hbI?m^5
zyn+uNvy!^#ZbriL3s`8I!B^ij#3HpfsG1}g5I0J=W0M~?Dvjs$Cm&(vi6m4il*yi&
zJfeaa!ADxV1aAfvB69p|XekoS_xObN^Ci?^?@5WLPLjqJeQMdDnj<&mCK|uSu?dq_
zGq0*QI5o%%qcXQZb=F(1ve=y*-n!B5_B;wp-ARVYPNdySmw#}4O{1+uuYF<%agX<C
zIp&XZx3`mx=yA)xi^mApI$ATs0B`2$B4gYt!E)<KMWTE1>DDO3ZJtTaWd^jfKjGuK
zOB8If2S;9d;#~6$QfTWb_5FPh=^nyAY%D=y+G5(XK7(yjSdZO76|8Hwccf!qPQMa!
z*r_dg^!dwO^mw-g(PHNQI9bdy4&Ejz*hHOgZ=xY_)3cQ$7fX8d|G*CROkzv!JK$ZG
z_?|Saf$oOW2>h5vTcQ@Cp-TqS+q+-*4*ODV?@CM%v*?GD#os5VHDq8o20S?z^)sK5
z?nWbA40WTDTZbiER{0{keKsk4oI~kr{&62~DW5od8H|0}nEba?KC7ydZx`K(B^}<-
z7?Dc;nIYsd>M!e+A^5e@WI9x_5$W0aWVkJbk5=4HDIu36YZ?#nIYVPuKvoB=uY2N^
zX*Y};JQfEUPvTO`URv|$4MwEq@+(0DF*Wir?6!K-g1^qZdFKbVa?eaUD(g;hf-Sbh
zqCY#W_KVzh%F(Vq_o4UApI4`6!{}Hj7T*zGp|#?Cb+{Ae9z8>krQUQ<L7(DxJ<ZzC
zW<@db#^|b9gp6~4$TzK8WWxe5-%0HGJ?7)<Yy;FC(1G>Dk34g33A^U}oLf2DaEl+^
zsMjbnM5%wL7ySyEmPsNOUwFjMW>3S(C+?^`wouad`wGx`AK3ssA+(>|2cs1?AyI8K
z-k!fhTaDcLt@O?K5@JPty9Eh_$_>N~`pCa7^u%AYvlLzsjjVZPl5Xm!vEa;fW;rGv
zwI&+EgLMQWCvU^CrZ@bH$`#h6XhEavzCkC-fodi_M*O@x*?T(gr#?&ju;^zqC^73g
z+jglC!$#*I`Lh{{-aVnENBZ(nEq!2dM4f)+DM`0A$k6~KAXk5j(X3$`asO=yX0Mco
z37btnw?AO2cjR*vFU*!Gu2>DH-W#E4q9N5<rX^7mIiwNuw?cQG72<xUVWG}0oO3kC
z@w?rap~+SJOTLZ0t7Nh<m(^)!k6AP{BM3ik+R^r0Z;aCGPX)&17}HcoFXma$?k*uv
ze9@ntip<Q;=Ju>&RTrAYiD`5`i))%6@$SG^k_6kqBG*yU#pV$7^>kzlzU*L{W+m*O
z<s2ln%gAX-IiGud5Y^p%N3Yfg^N4Xfsc~}@JG?W75{9nCwKEZT+VeCyWTcUmD7Rj@
zv<=4YdF0aPuwiXS2W9@~DQz-dPsg%)LCbLuwk8s{3qGDb=twNdH7HB+RTXn|EyV0D
zED#fCT;vOfgup;OO;YmF8dLFv8}DkM(sgABoc9Q}x{9#+?SV_X`#~)|Sf+9PBvx7_
zLGq`LYK8_w+iWT|n^r=mehsr+Ea-HO9c?Mfp>qSujMkQQ&dH0IL7oTmd9p>REIdo}
zZV&dt1XnHUX;;e=ExnM}+e@Z8Jc3sQ<RW61=q4)#U`dM)9`xBlYMN%4@LZ2NGfOu8
z&r16BJ&UW9oZz#}jnH?J__kfk!7sBqcIrkI^<4QM<P?u%##bBokMg020XEoq;RywY
zt(PedazmCwD*vbcL->`0QQJ11@^WYMw?hKyzN^@yR)*u)o9nDwn;je`Zsk);6mW9b
zJ=wpti85UlNy=B>a9dwLw#>AGI(J?G=dXpL6LJPgJyO}apJpVx9>xFfm3BJw1^Rs#
zBlN#gvcBuc_g~NA=Y6+OW|v0c-Yr0Sz7uLfN5W)BZ~FU9N!r-`F~*HlKuZ5uG_H;0
za&ruEX<QtQe7KmqySia%_;VKeOn9DCmtvS8dz!`m;{T$(5ia?_mS88DmKMv@cnj(F
z_(o@26w!UNHhpikl#II&!%aozqhr|$G-|ZcMAg?kFEAR~TX)MA$-?kmaUdp`naZRl
zUC1Lgi#7VrB#FvvymHn=%^n{L_p9a-_W`VQ+A6pXF5yS~0_j`W7PkLJ6`TIj9>1qO
zrjM;B(00x>>wNoXe*Hux4?5_LLy|lk>ZvGs|3s0_&NZRS?VWJ@i!-*5P|a!G=Z@pI
zGl3!U{NH=UoCRBwxqeU~%nau78+VIY!Ovc#S>z9eQ@vQ5#8R?<fIBAr6SI2<H_0}G
zi&!W(OgMDB>AIF9wu*Cnf#-CpQWbZW5B1phHw-%qJ=vgbC*aXh!$<h<=E{0=1^2EJ
zcfv|!Gyk4voy{ll+FA|4uHPrQ^rM_IOzz0eg?!_4cF)7swMjT9Z-yb?US{vsc!-X;
zyJ(EzOilk8Ili(G_lp>$WI4-f{QKa6(jV%xP7}|~Vo5jTy6mZnA@v<CIL@Mzx<_v{
z+O&Vm&Zw49&4oM+QCv)F=Y-qwYaJ?kd|@xMPNHr7QCh#IiODaxPW{&YmRu2C0nL9i
zpmsR}XZn|;@I(fCs(BK_WDCfx(*s1<J))s1ZaDi=bet#NmUZ17i1XKEFl?6RbFQhv
zUey_6lC0SG9#=4^Lk=#b^6XkloUB0gilpnMc_KGv$~tuuXQn$2q;x)pzqGffkw?yv
z!l%JV|KvvPS584&-$T+**_}EqkkS6@yCC=e8+LoFg537eNIU=0s6E`6$*YQ9TSzPG
zX6uVl{3NYX3L!b2{%lC#E?R$}R<dcE4i25G=hg$nOsY|fewyX-6Y>4&$A0m5U8jR>
zF?v8_drzc0%5u_)8c7s5I)d+=o)5=Q!T9&L0nf74X~3Q;*_H*H(OQ3$mRI^os-JDb
zzdUD@&J_&B0nDi4=5Aa)8%|3FpX^>nHS~19qf|A8H5}6xcl|6>zSO{@lcD(9<p$E$
zrC`YK9GP`}8&~;W$18N>aa}Ojlq7ywd!m@VAMH*5?0rb)B#+>_aFOMG&&)*z?Th^d
zTBJT4NAqU$SuZ!R`8PzLBki%^JN;$$f!EM@?F!!P?40vTD-UVI#!^~KCRRN+hN`T9
z(w>T3R;U+U8MTZpbFF5Hnsec5E9PDo(u9BeDpYSNk*c*bjWzX$+Ol&zHTgU(YATl8
z@XD49>GB>}*q^=Id51r}dX1#pHAQys1pR!VB8#nwM1@@%CTU(`i=rd&+r9u^590As
zoWU>rF-Pao&-t77dnnKi!w}197@+zL%2P@Z@-&v(r}f3kiUPa}oDJ_aQ@Dx9$!V;=
ziMsH=sQWgT+P>KE*L`JVulfaZe8Q=E<P)^{Fq9QXV5;UbRNuV=iXKMQ-OoXLwJ*DU
zIu%Y8AJ~my(Snz;8iJgK?a@QXVa;P)E1D1YzB@7cU?JOh`+{W6u6p6U(`FG}bNH^?
zTiJ%Tfr!u)PUx%#mfvF%vN{;862Ies=>_!nnHgSukESU~9>^_oMXc^MnThXQGP5*c
z_Xhrgp}ewmRGpXbjpb6g{Q^iP?;%s)%{W=<j$p|Z-1aUZ&zG54JVO^=V&@P!aXguQ
zamD&#1!;M17F4fW!m*<s)t<prJ0dGv$53$Q_Kd{93ARYscoG-Sws8Z|O)SVe3aw85
z(5@(8W+Pugx$Ztc`%>hiS1uJDlj(ffkW(0A^@?n@pOc}^cD&kj4l4#tqB+l81P^yM
z`%t|D^!ujF?6n-W-}^?o>MFEw_H3^DYdHE`43sTevWSjPk?~xWk!(Yk6g->m$M%jk
zr%dS$iWalFfVm;0Qj|dR_VpuGTLscHGRED9-Za{xmLSfwyRF>$q@i1o;ugZyHI~ub
zTaRf}#aupXke88_Ef+Jk!xGt^8nXYLjltVaGP7;#plH;adf4fc&AN@W)xR44$A}g8
zd6NC=UIaZ-|G-aMo~EI%MsQq+#kFS&)YIdStS-9~E4p%q?CUb<!<0_a&?z2t>5dY7
zs|u04rXaiA;BvOk?^|daT#1T=#gw*b8oHIXpiijCm$#iG=c@ITwm5-iDxHE}muJ{l
z7D{PDR`Dg{t@w8rP2@xsB16rATB;jl;k8P5_`sGdYxmKJF(vH9_U&*eI?XaZ7{lAZ
zlooDv!u8q)bRE153I3YYO*}J(-$x)gTZVzd?$RW=V5H6b%PRW0QS_}BvdyXy*mi6w
zPndg@J>S<4<?}AXzS@IL?XncI&^WS~WWkRvy@Q)ObtyB)n0_tW&eoN+(L^yTj(*gG
z?RZv11#Mok*N%GFT|=ar`I8@#DpG0HRmweYi^*EyLh#@x_&Fb0yW$4^SoaHFwr+*T
zoMdFhTtU>ur6`;`itOr0*8R^|*&ERf)YUr8>TwLVdj2?F=E+U`4Y2dZ4#CA};sb5p
zu)Up5A|$Gi>6GSS>%bP7<Ek$(I$=y3lAkbRcX`?S?^|S9C%m!KP}^vEuN=fisz^P|
zgbTza6iZ|&&@~pE(4=ALJU<_kI(4vvg7>jf%r;zy7UJ@yFy11!7sG#sOYVx<#m+9t
z^iuFF*L7b?$tA^<u6By#SB$2Zmg&%*KN-Vfc46LRH`XTJ?e9B75aZ^9!RLf)wxNz!
zj*H@*Ps?M*mqKp;;(}lvU8b!0w~#Pv7U?-w!80$B{5{XmfFs^Ccbz=kH(kT$n4yR`
zuSQ!=xbxOe;Z)4UJm!rWz4lLsRA&|$gh*tsZQtNu-$j_z?>WBgd4w*v()soF#Te?Z
zL>h-LNgOpMAas=$Ux)d$*07Q7+B}j*?B7d{RyVO#*McQfsFD4EL>f`k2h%dvbKftc
zm{(69gkN;VGS41#TD+G7`z0akQYt=YU1a*VJ~KZzb$IL_O1;tp@b<rBFev_r(|LaM
z%sv9QV{&MP-73moA0b@5w*1o7e$?4#ApPnQ#B%bLVLa0*`&|A>ayADI3%boJ4hQ1U
zXHA;@^Cop$`va2-uA%VaC~O<Yxbv5#B=_aBY->#q@jXt0BfE~jo!aSqmEf_q4yPCI
z0!aItaBEg+ONP7kquM<NG~NF<0@VIU`nUC`Wxb8C^{^9q*NuV0JtxTWOeGyHg|xYG
zIHIcWlG=ho8ro+NZ;(!-*wg3ee#$&N-K7u9>mm={mkDo;G3+#6%Fc}m#_yv>?3Mj|
z_CcqF*B_Bkz{D^3S|_mT*N#!#<=^bh_ak(ws2yYWPotQPGjV?WSYH4BCGt8?#+1zu
zC^B{qrg`3_3DQ6e@e-u;CwgM1befW_H?s2fo}~QdDsy$YLO#p3!fE~~)OSrqjKL@#
z_4qe7JbHzvJQ{_=g`?CYmy$Pn(Yo^4xRq7J3tnns-srB(t%kA1$G_0KAX{8ooQ(~6
zw|Tr3r^yu$NYqW?q<LO8Ot*!G)D6h~6%YrHmz#O#&!Srv;EU1UyC87fBb=}h9-*--
zX>mXo>B$Xixz#8&>99L~xNSCt?;86a0}d2&g=rzwG$k6+9>FYs`#4Guf6r^RLn(1-
zUe>_}U8Fnxqv53^cAU)x$ZzXS>G3)=x%e@3NB2VXoqz1nhtAYHWIt^fS4O*+B=dL4
z20Wq1Zsut|nCxRZ@F=#6bngB@)Lwlcdv36neNcOd`QFt`v85k8XXIksAO++E|A9pG
zCf4?-=JFwvxLY4XkypP!m-_ZafRea#?|RNGp6?*ZG#51QZGgkN4B?eKL1VUYo}S!-
ztId)0<>Fc}!}VxOZNe4l6>RTz3^$z4qHoa^I)0}F_lFIZwFP#kYH0yki=Xw5iC{_P
z-T0uMuPE)fH*c=pLW||M!mQ37x!IAjU#83Gh3-YZZAKc!cb&_}{b}V6E05q&#ytdV
z6kf3D-Za5z2lf7YAD0H)<13`9?DoI{k#`NE!@n-@ry4`B<ddr;dJ-3Y>w2u17lOV$
z#kpykG26SffvLx=#=CPXusGo&>sZ=C#p8ampaI$VqC5)|+GSjBkOAJ$)8GT`bXfj)
z6{*RzNV;Zx01k@`kkj;--br3#i_0_m9kr7VoZ3q>9oLZG#NN0!#1BIf?_$RR(K8+_
zn4ndiP`m#;+Ysu`((RtI1!2!IxLiu*#(&uP5w%7h23`3QxmyTv>x?1GKd^~+*Reix
z_VZ&OLztUV8W~r3q35_Vu+Jv&-DFFr$M~c2q&n?d-<98Y@G)w6`#@r|&w}jlKjEex
z@lg4@0*=<oh(Bab?X^)@w$>lbo9m2R8*>GZ;Uvp<Y{J7yMK~juhy732Q0eN`Y~=1{
z*w2t*hvyu!Zck&s{%aK(jB5NFU(dDI+QHeQh)-9x5)O#}aIsm1b);~VUZ`f36E0xH
zPFq@#DE#%3wfOO75gL}aQ0|}I&`bhG{Ef$w&dn$^4M5<JcsN`h$g`5BqjIk=a)!FV
zQgI#z)yiSL)E$~{o)dRy$FbmNm{MU#{D&cCzE`2?V)m=%m4Kc64UTyS(N?8ssyn)d
zwWRc6Dj$5gjwwf8(`VVD?ys5cJb7poMF}UsZb{oabJn0BxW_Y8rJqgf_yA*V>a%qW
z4En$4Kj}Y&1?usCd-JhR$C-A2m`0u4bRbzPaxf=TF?8Z_y7)1IOdeiB(6iZ)VLaVc
zRfbjDarR5)H>a^LxO2fUl7H^XmSu}<%|mq>m!2j`+IK}FLS>LE6-;Io3w}yIk4Fy-
zMW*I8S#v+Zo>lWh+t%;sKDJ3R^6W^FTdhR+&m7k6;auDu97yDRAF*Cy_9?z!Mj<w+
z>QT<Njrqqm*L0<F^_hr>s9*uzMYn#_Zl*gT8}n1_N#|KP(hj7N+p7zhYMD<97A%EH
z#0zE@7zY3OK4`r%nkt<#$!ft>di<-Cw3FXUl+Nyw<2@{x+l+Vs9Rm}7e}M-YM=DXp
z;5}Tkr4Vn1KfwN+@zj=VPQOGqTH92W{q=f|jJkZuiw(!QNuaM_6<Sf4O*P_reqj2W
zPQr9@H_5E;-}$B<9P>WR=E2h)u<(;`XkOe8*GX5&?bB(RWxh>t7?-g(JtsnIxA;a3
z?uCG7t-LV!m#oy}JleB0k@(;&Gc$Ffy=Sw<nfN<Y(nsQKNEhsyXMl4nqM6}XbGEKu
z873UwP22B<k>xl8I7G~)F?QWy<$8r}-7}WthR5U9MZwJf=asGdZ~_13T7Z>?)2Sg|
zuz-BVP?YvK{(fB!Ip&Xm?}qERdixmZNlxS9f%E+9Jaaty(wi^2J`jBZo*20p_cKJF
zE09i0z{xpo{OXE)-1wI+_yWavHYN+EQ3qxHT^niL#n;)}$HsDJmr6XUT!@4B$D`ki
zx1`-}Or28Q$WCjX(fTdl;CRTMzZ3A5K$VlYQ;-VN^lEJF=ue|G-$J>Jv+ncaWQtwY
zbEI!Wg+Hc<^?W>*>5u%1hacv175&q^wfzAPEfF2r&JAQ)ZpUA~55>5t9Z1`i%KR_S
z!Qr1z*`;@};Gbex^!s!gta+8KEOw%m%k=q^pzf^e;$r^7Jd|1&%EUKkCHZys;~pDB
zk?z`ynr&Oz`WMRRe*X)PpZFi!cl|#MaU2TS(lqG(3B?U^gsRwC|8O@oQW+eAofl8Z
z4qw`gfO1tF`YL$Nzs54-X+7w6{tw)@a-+YF(fC?f$*$jzqm|t|NqdFI!>9U`<Ux+$
zIjekxtJob|9~3$5>8?=dSxmcT)iH-+Z`vJ`%F{2dph<cqc#)EW!KI5aysH7mCqBW{
zBJn+1FiCQ8<6<6d5``4eS<DL4l1+Z7!{cuzz@{b;H=<mT`PrDmrHNX5ak{N<2{YR|
zY;NdJwRYKvi)x^L6TiVL(-W>ghO={vG7vEAhVY5kvQ>WlvF+%0-qH|3#a6X6uzfGz
zZxF)Ud&lEs<9ck_cawI%m`QO?arkpaUV2*aKAfBSBBk{y>fW3~{N9O1^GrA3poOC3
z!8u)=?SBTRM~){`cV#?MIZ1!|%S#ujXUH;Ndy#Q^7FLAGxVKIYdpA~D>axs|`7%$I
zVBCn@ATPtWw?5Om#KGihc7p47s(^dD8<%_c62bAol=vkDb|sP7scSkjmot&v_T^X9
zj5L##ckgXzwNxf692W`s*gs6YcQ`#*mM*E>@rx-8Rmd4Y-6`OKAB`@W4t~&v><*P+
zwA(w>;V+YmYccH6`#l}ZSEpH}huI=tL&wIH0GFGyzl`Z%dB$_tuIwg0^~*V$WpE2_
zr%cdw%u^cj!WMZ|fjn^63rzVJ$P7dK!+A&sE+0!{-M0lGLB9(>y4(r{iYIBcdOB>&
zcH&I>S@uzMq~xmO;i9!Adw7I1_j-R-FtjWAXSEdepYvV5+&7<D@vH1q<O)Qmr}Mq{
z<)JPyl~^rrmyH|lLFX4-;4KU6B{gT{X-fACvLP2eXveXqY)Yz<=nc(61vz2RryFca
zc?C^yN~TNwZ`0rMyZk@ZRdi93BEEC(WTRw?Nl`!XCdm<J4C<)YU0_#pt>kh-h-}4i
zOWF5lTPek+2d*3m;|uoOLTq&eTwD5aW4j3ma&E>L<1)T+WCYiW+#zx7F2ldo-FR}S
zH!{|iiVRRJwSG~hf%SgOz1<s+8g9Wrdm0@aWh)zf!yMo9)>3A83-1;>kR4oULUJ2_
zK-oHo40C-kqFfn=JaTDLnUCO~Z(<D}%3)I7iOg#n=+~q9IM;mx)hopzf2cX9o5H&}
zFAN7xg=0z&FBIQMW-~(Hvjvmv$gOKP`mo`-Y{2_9*!;;!I8tBpOg}%AE9zs9^aOT4
zcu3w~%=rn+NhJGGlU-mY7&K3|<La?k96e!8-`l$87~GzWYd%|9d--CPbLl$dMQ6QA
z(UG4r-Ge~S!_fV4k=g=Eus_%mJ+<S}*wcbddi;h8r+bmsst*48_yyLh%ANXLR)wtc
zJGHDmOU*-%k)vl274LY#bcBoIOYKE&do&$uCNnsmyv%fVP38xJqR3jC^IqmXFt)BU
z)Ek5|%yTS7cPt=TiZvb?g|IzuH_-Tguef@K25gv0&Z>Je5Vbv=M)mrS{9M*EjT`2?
z<WLfJ#*3W7rMdjSuSz(t7mtl2@8F%YHc3`>$~l!*N-C?4QHr#cMI@|dL)?b5CHH>I
zx>~xUOYmVba*U$tJA<ix_YVGG_)sVw8-?_Or6R|1kXe1HW`m9@lkJf@vQ@9+cjv4j
z-9Q^$-pQyfyGatZwGH1+nBiM;1xvD(XNz_hv3X1N=-!qbY;5-7N46VbWTOjx)-*Pn
zfBdBAJ^jFuzV|?j;^;?6C~5fIfL^mI)cZM0qB;d&PO1)q7mlP2W3x$ZZ5mk~4X2K0
zO48W;S{$wv@3zTM>X&7M7w&$%ztM4?o&ErKlEZ2G_=nikq%4>KalE0$fW<9KW)3x$
z7}k(Z3R^GZx^f{bZzf`eZ7$p7TaC)JGuT^{jc>E=(}pp9CDy)n&|I$0n%|mBv_td}
zDGj8IPMMT7se4YC>lZj2@t|P?Gg<R!JK@UxE#qc_pOI*TUAxkGn|T2pH4s^lQyKhg
z@fW6MVn+G~*;v#q62+HWjnp?LaQlCjY~9B3Oy4`42kbphPZuPk$-;&@>(o<}V9BTt
z491iYdz|~dmxq6NB3pRuH$42?Q4s_*Y+FdLKCFh7`f>`me4pAZKeDy|1tL62McOWU
zs7Iys=)J@i%6(^YyVyJ)?D3PPXqQ1_-Zq4ozGafekFc;m0&kTWSf*!&v5o%pMf*4k
z`pn_Ro>wKMhD)$t{Q^(uB!_G7iutOpp^}=RrZ{)cRid_eisVn(A)x9unzTI0NTZl6
z(raO_70g^eXY$&D<J9NqDZCoE6$3LPsA^#z>*MW7pN@FTBv0P)UKeh$@dH3UQGrzZ
zB7k)FZ9}(!t5{@y568?;QGu@ua<1s|n5r8jdF(E0DgA>fpKN*YwPoz{tOckY=8C^l
zukpzh78oXadCiBbaM<I%aLM08Tt+80x1)hOD{jN`_6dC62Y<*Xy(78oVs`YdIg(rJ
z`L07^hj}5E__ygOH}_@^bJrSeZS6;K6>jwKvXWHVdl34iPsGSof0^YFchvsejyY34
z%9IWlOA^%+Mc=0r25u{5igX2=WZSVQbqAf(_rQm*kHoj-C(}B49sx`0@Mn;@q)@8K
zPQUxh#?QEgUU`{teSVD<^*JjEU%H50?2^fnZ?2-Ui52XV_at`AW(^ezZh7xRGTGWu
zlj+L8dkELqg{c3OxQV$MGM<nudfPPmvB^<3`_}^~e@jPkzZHDo*>oHbZbkiBN@z9@
zmzga|XFGo?&|mKvvUk=ic(-ZV$RBYEdv5IFb4Sghi^aR~xo#uf8TOM8@vhHGDe^Mf
zU3!ORKPn+hkNNbiaR~)S<YBpdIQq2SLCj4pbiZMYU#XtLlk%EgcHHK_#y+LZ8)rdr
z?+#SnXlBWB2QWKBS$fp(A^o*AraKo$vYp#=5puRSd^=ad=Z-$CRf=HTH`)kyH}TbT
z!gxXVb)tIdkFX1U+0&^7RQ+`(CZH#^&vVD6$di(ZqwV36P(j_7$Yl2E3XsjMrkjnu
zS-im-oOe#-u4S?GQqdd>w}#TZABy053%F|3SiHElQ+N&*kmH6>ZW#ZM2MXQM*Pv)<
z{ktvso?%RHCzlGI;7Sa3{KE#{FXpvYx)eS$nBRY=CiP!ZC|fK#*H%f3U@@4G_Qnk}
zonG+d{7W?ORt&2<(S@y?;m&^@*+p+|D^ljU<ZQXU1JTcI9Vt0Vaq(^j4teL()`Bp^
zog6Lu(ThkNY>@f?-D_1nGPBa?V?qOR^UkxS<3%^yO)wf0&(Z{)L+skWrC43HQFd=?
zAkIELPM%epso_Zp(nAH~#_AeVasL7T8QL`9c@v-UA^;9K19+Hlg6vA2#pPYprS^B{
z(8$*t(cNMmd_NWA^%}ux%~wHC`+k<x=8s-i{BXr{IL^o3qvW^(nqg}~my-G++&=@J
znd@OF2|><eXFTt+m~`BC!8gVLOTIs)zP)c^(%I2``xqxST=YN_hVR9}#1W!5JqacS
zWr82khUR>KI&}09IqQ9v)%MB4i@52SGdPmMoL+I+!a%CAG2+W@A7_7faEj&TDH*;0
zn}(mmKd__!o>RrRBm6+G6V!6r1$7@(Nwx4aY(B)Jr&L+`)Z`Euuae`sTMuJ&=|r%F
zOJINU7*gJxBg?GQ#6SLnT$ez^oH~V_-Bg(Ak2}!+xBzoD&O?^ilNIH+(CLqV_|5^R
z$n}>6{l4f)>bfNeo7IaR_`2hpNjUC|mgkQ;71Pf^MHC5Fg=_SBKBwZPggtGeUt=an
zW~VUL;v@RDIa8@;mwNnE9ZJ?;%iwhXHNs@`DBvfc*Pumvh5PO9(3j#{rAEUhOIalA
zB>6J_6dj!Ti?6X%l^PiDq@ZP!nVxDk6&Q}i)D<p*2R;=IZNkxBvyV!B{P|}6i8AX$
zyD53AJH6iA7iEPWSZ5uEU7EJ!vaTy+Ik6It(y2K9BbOJdUFChA&Be(s@-Q?`qC7Ku
z3R$*}>%T0A)8fuhXt76)w;U}vXUU>X2jH&gUTWKPM(LS(BI6s%=+ID9`wEtP=@A^U
zbwlmrN_=^$fjcTeu&KO3*Qc3~yvu!Bm9-RmXMds*YN?oXyP8y+46*XYU?zY3zN~F~
z6m1Q&!~TLBFo@rb*w?{aY0+H98#V9&PuM8G9oYH#ywMF=2OCuOnWZZWp1|X9ERUUl
zCC<Ys;(#q3{@_J>=lM~JZ?J53awtUNnkIZOA;*e1d|#;|Yc8KAx|y>OBxWK{KRQWL
z)CkKfqhYa02|Dkmu*O6WgzTI|qrEF-UzgtHTNhiwRMm-!jKw|PC>g<rGAX#v2f-8d
z7x$@ZzC<wJ`}KLiTV~a>Y2BV7@Ax64M1=~*d?-ftI)%%*HX?f#jL!0X@nxGXCLEu~
z*6SD1-OQU@^4*P$A5F&ol6M%g@-^Q*CkF|lL#)^IiJJ4nkgC-~df*3P)t+2BoMg_f
zFE${}`Q>c!U@s;c+k-ZRjlx3<6Lh?Dk<3qBfd3Aip`UT76j)x(TlEx?X_Zc$BO>_c
z-s!C4LKOA<87tehPl@rp2}l`T!swuuB(gA3VldE?+#e6d;@}8`Ri6^9`#f2vz$iYt
zpEgYscRtlqQ)u4MNH%8tRr;!MN@jfFEPPW3@;)P{3y)bp_WY0+ZZ+;f(fh@0mC|`y
zUE|Hi9q>h>)?%FM9FNeV7u<S_J|@(s8ubkG##8sdY{Bx*jE+CTmdQRSFS|xJL+d$K
z^p(tfsGcKHKEMvSzT-;<1W@lS@s!>z1`GKa!7yASnbPAht=e7!gNC_WF`;Wtv(0;1
zvgQK#SbM?TeIi;O|B(!`_NS0}Zj^QCCaE2Y!->A8)G>7{ztdYzntjZj!d>cQ9`5dF
z+7LodKgr-SUN?(nsA76&-|Rk0anvt9m+6?bv*}-m|I<ukrHOlxail-I@)dJ7I$D#a
zWgpUc`yM%FFW|N3A`RUos>;^Ua5^=Nrk7mL7KDEqeO&=PSD0X)V-AlreGhe=^E}np
zjmr+IiOgprRjgNIf5(YC<g{G+e&(#;?mIKn;xcqExQ}jUGLg39D&`LO%v<G`z)x_8
z_RHq59ILZ}8+`{M*-0dScnAC1;Y^z|{_rV}2e8b|*RcIt4J~>)QFirYk*t4<c*oV4
z>}17S{94m22{YORqpwq_cziRo;&&U}Qw^iuO(#STU{toSDNy<57tsIYNv$Gxs5-fo
ze|KKa%#DX*N`7Zal+AUL_h%TWUdiFA$r^Gu^G(GMu}J4QEmVlbBC}A)`Zl1Gsw+#m
z8b=#u{9`8;IH6)>H;fL~U@{G{x1DndbDf{#wNEqCX6`s_>5G`)C{py1aku?yIsT&e
zT&Fe>;}7-2uCDSq{R{@+onjzg)EX_p?+O$xTf&Z=ET*f*3e2ZI7d4xtWZ(NO$%QoC
zI)C{llC<owZ$&i1Z)>stWW)IRnZo}Rybt>>yHRR_5sWPk(%flUv|Gm&Q!bTZ#wZQs
z{;QxLlW(ABej<#jrX#1U0ci$H;QDx)IO_*v`j(wkWfTusXY-Lo$q<5DaLsy_RFlqa
zf>*KM<t@y%-*EPIMkpz}d0^o~(c>~wk{&P|%ex0gQpm{zs8?{qhkxNLeCkBnB)5|K
zJpae-8=mp1SGD-OU?>0EQw#t4jH9u4>{v`$7J2;)hshVggbg<)>*>nUJ8go8WwTl)
zcjqWHGP}~#7#$iYcZWW_2}1DOsr3Dp0@NR$rF(Ac_-ox$yl7?#C9E*Q1pf<kwZ~mq
z=ZnH=y|0paTs}xij`BHfHU-)FCzo;Gc`lO97tQEE>pz_Mae|%cUCNGa@@0D}otR%n
zA{L#Pg*hv;=*W_<U_&3k@?S7D4xfb$<q7<f`y1BW_cS$NJbGk&X63q}lyl67=8cP{
z>R(++8h;d-vfo%WJQM@Rynvci9&hx74{doj_C-bbb)NdbIM)P2W?YiErgl(}o&~S@
zub5X3a^NGLerCSzUV<B!N;TStso_s8%@Ewhu(gJGZ)y$4>ID3!?9V=WU8LK8tJ(DG
z7dUy?4WACahuNhkP|vNx=s$`%>-00Jc%UIZ-jva-9p0o{aFBh|OoPi4d2x=F@cNO9
zB$wwW;k=xtQR|4gXb8W7Ze8SZmRNSDW}gCnd+8_E=UfI}7w3`vZovTUxJ$;<YT)LO
z$P&NS;PKuim=H6Njq+3$-n48obCRPEapCCNRDk-15a_z)iT>O*R1Fjy*vVIrav}nD
zuSNH=QzPFTsLeMNH^>~soa*fkWwIaDTT&Y(X29)3sQZ&-o-wKv>!u7xTY@-CoiU)E
z4fV27^_Qsa`b~bK*arjGcapZ*Zp5ZeQBWRc3PpnnSY@a}a(fLad~_<wJ;+1zlo0Yy
z%Hq$qwXu%F+tIC?1wUx%fmti(lhdXPWVq@Y+V(x<f9r1Wi5KPh4}C>^da#Ki&XrM`
z#$qbilL1&<6*F~1YJ9X$u!ZC3Xpk=(`7n&XYmz62y&aOZBZoq3Qo5w%LMGSo?uO+9
zD>3F`Jl+2=f&UwFimN}fM^lF$B{Vj`GWRcYXf9w%7Bg_X`%J1EE2VvXouTU~m3*Tn
z6ii5gbAFL5ZiyZl8^&e(u2saE3wto*-&xEq8O=@mroeJ}FSh6OQX0avDJ$8Zq<aMK
z=(in<%3F>wiwvfIE|La63jn#4BirdV*nM>zd}2*E4<%7Tv|w4^Dda~7cu;p$b2jPL
zFVVr!!H(}ga5UT|+ePsEW=(NqiD6CrT=G^*Zj8c&MNe6+?H$R{)py8vz+t*HyBm4b
zxNx~HR^&SJ2Ev5<AbU@$;NI>=_n3FGx7C*9JW-Ll?)xg!tW>1h-CdEpTpK60rtrO!
zZqaxO#IF4=*w;N5R+Tx}dS^JyQyIF)gO#?P6<+deWSIBlRx`KI^5PA$U8g6q30qPy
zD*7aE>UIdj?VrI|f0p2=JV)>WfBM){muCLU#+K)?Ebgrl8O7bFKG}mQBBqTmfiF{b
zzr}WVc_Lf#g3*rq@Q@Q(HEQ5?OU_}h{7u8;Bav7cTfs+;?T_ab3-Q(YF`~N3Xv!^r
z7H;<lZ$?|e^;9VjnjOOn@Bd~4<ZPMtUUxdE>V(UFn`q@|JvQODIp6)=oQBx$WrGwA
z@VxpFf-Rb`+3TQ8((f*PAFlwjeOa`lhdu^eKEWK+`tbe<O9UreM^yK|ket)9thSvl
z*+WK3CAzXE?j5?7hOfBE;>F&?_jEp;5ATgWot>z==tA+;J86aPezF|6g{_L&0ktu`
z5WB>K(!|~2T5~y$X20WWO#(1Zwi_pu?P-Gjd2(1g2t%KwW9`sqIMXKP<XazNtYE92
zi%sKQPwMe|ja9Oa)>|wtBwM02=@`k1Evdw1F@<R~(c8G~XkHV+r<(iV>Y+6h@AQ=2
z&T<hRt9!WjNRE$-(}2U_Y4H3{Pm*4@9#M8Fa2c(`-rn8-i}ueb?w=@`)3TNwbQ(`p
zTE1jo<VXDBM(As+)7tg!aGAdt$6n?_GGrQDH#@*Q&jsIel(DGObJ$pwu>W!jXiUUn
z^ekA;{`l%+<dbVOa{C&7`9wNW_WogU2b`()f;a5Sp5kfvQLG7bXDN~Uam)8SX+@mo
zp{p)pLWbA}Y<&jT&o?DH?6gGsdn~h>bdRRHtV556c)oh97FIlqMe26W>bt$i6R9&L
z42i{(B_~O*_z^Gf{(?^1+VeGoL)hVh$5hqkO#|9*Axf6T;W3*gs_$jicOS{Zvd_`)
z)#ivcZ<XaQ=_l!^u40c@-QXYXud)Gczv-)k9kx|yU}p1498wbgo7FuezYW}IjhLIx
zQCuS#WPBZyRQl2e-CWiz^XCp?R&eO+d(!PW70Y(Tu!-6Vv=A4t-ti&sn`iOg@2^n6
zZRISvY3aDJ;{X~oyGT1NUxCl(yVBbWg~%G_PXqF`_`CCCaI7+&wB}qxL8kCzxQqFD
z&S=a#oF`j%w~gy=xWZp#cEW@iK1gg<mad9Az)RIC=$PLnwsKitzG|r{y_$D{KOFUd
ztv=$y^PPmVEh>pjgVtc@(B;^o97da8&t`L4L%FV`mDArt;M(a3z<*>mR`7id+F7Yj
zJuAC^1C7h;SW5a-bd;xK_>*8Nekx{@^1eoAcU@yo4?kmX@B7l#-rh7s{jzLzsUq1Z
zNa%FQNTgkv%&)t8GwHWr410QvuiIEgrElY5x+@)#Mk@T*`drDv*j9RI;voy_k%YoU
zJ80IdLP@LoIlQ~)fhi3E=;u0IHu3E^Olxvw=UOslA8PZ-QEwT^y^AEDD>hX1^&q7L
z1)$_sICash#zn8)!e`$~ct9+~cLBEewo=6cktDkRh$R;X<K1;zZ2t2K=b}AvUoH}p
zm;Ht0%s~{FX|TQG`SP3n1RvaW$^4n3G@r8h!zXo;O&*!tWWobxeJ_VC>28k#c7kWC
z&m=dc5*+!w3tOKzVUm0n?5#4fsHGh(Pj1oh7AZFLh=sbFKO$$2=P}CPm`69kr_Z~F
z$Ul>037yJhOUElv(7>*;F&>jyo~<g)P`*c2)#0QxPm7lYW#V2~D)nETiLO3PtnQ%=
z+dcX+sbw`&s_@VLU0T8GU1u5n3H}fHleO{d%N8iC3B>N!aP+?9NB51N$)@`x^9_Zc
zg%5ot?d&!g$^8RpYMdv#H~AR-5WC}J{{4CR-_>+OqDqgC#$s~C5zJpP2fu&sp<wG=
zHm*D$y?>5G`W4|i8dXMm2G1n!?~=%2Fo*dr4LWRfLbz@&@rppPPd@S%w^CoBBJ>81
zJ{*L$ej1d0;~(iiN#pVry(#lqHnWd6rnM7<yV+lc-&Rw}$fqv_{xu=gKF77I!ew(b
z3CRJ1Wm?@y`sV8!SXherbHEFUMtuOkW_^=pjVos>+YFhW(HTrp4MNzF8gB7539H<j
zvEj*THu;YqB_Gt`wjaD{kTT$-l!}DGFOah12Qw24OU+jre4METZ`<QZ#urNH($sIT
zWg1wk6G&yDt7zTE8kn);2!8b&r)rw;^225L<=>Poc$I~Bt7JGS>mofot(JE8zD%ck
zmuEGrL`uGTpQgHr^Pm{%fynv9y|*uCrG3-!PktBrYI@R=#7rto)<%Qz0eZG%5)x)i
zrIqPDnAc?+*xN>kXU2*;Znlz^apxT4s<~N5R3rJ|$!mr0<Fe@18Nhq+LjG9Vg2pjJ
z_=MM|k(rnc&rM6=RhvLH+dbho$eolc8W{Np!eHeYSw(0*xm`$=WLJCOe90@}o{gc6
zf}3$;!UnOkyU5o02{&&_y)50TkcJDV+Ma9u(N&s3U*rw&?d5)c*Uy`}Hps)qeFxPK
z*Ff>#+n7963Ev-2m&L9b1G$%eGx}(q#<ZpC!tWd)eDjr(-MQKL6<kCsPk*37zx9IG
zaR7(O3cr`WCA+^5jijA)nY1bvm!idfaE=XrxEJF1vm9Jrb%4U2JvZ8x`4A@WmFPcp
zBeb8bWr2go@UP$g(D`O77XRlDtaskQa<vEKsC6BW-=?rgpOIKHHkuk=X;7BQ7?!87
z43dFkaN(dSR2Rj`#_Vb5gK7`blOqMBb@dco|2IJ7ccL*aXFO`2<gpKfg5b6GInS72
zhgUySWVXjlc#6Y4{_N^+o@^5&&hueNXxqn5D>t)sDjd(Wr=n3y2IqKXcxgO_J}o2r
z&`n5RSSedKCyPGLU5wz@1MqdQDoZ#(k~PvijMukf`}V2w+~p})s1%NXM^TWi`vD7|
zD->?3f{H(@c=6g*M(by8L!Xju2;6i5t5n}eS_C7pexD3#w4d*<nN8|#4oqLNP0Wp(
zBym;KWU;4Dip=B+_|#8l0Ylf)ojFFV>zo^S{?CpE&h;csI)ua0PEwV1Q`oUy;(h9(
zO2>k?(c>F>bTp@puX0|G;sfVdrG_Cj56k2oXOgI5ehUi5AK~dWqCY+63?|sMlFse9
zNSSgKF&%2OeDNi0_&tVRem#uU*IFs6UK5W8WKqP@65MbcO<rZ`2#gCu%yp5y(n;Y_
zLA|JPumWa^_b>eWZu;+l3;KW6rXxT4@~tlwr3sHXTeANq<kG!pUQh|wEEY4=ZsEwh
zZHJwv^V!Azb`&h~n4xdHD5+v1I&_PH<Kb+;MpfM2aGxdijNs>6e>3y0Sy(jmE>>q+
z!t<0d9<J$*zklY?_2*qN*2M@XToc3`?>hF|G~&e6eOQ^)7sGth5U}z+cc)x9+<3(P
zy7^(GX|hB+aTT6@>&`O&o5QoKzH!Ol@siFJ!_e>QZKH8T!m%rtPij(5&tnUy^-h13
zcfL-M4t`AiGHA&*IgI&UgS*Lc(!HW{Q4qG6lJ@3_u0}JBbkL*n;v#yk=t?&$l|W&=
zartpILi03v*LR$cDvG1?+hR%ELi8CY$5OcGM?NcU5&|du;?G{cr`VVOc%PZViEh3T
z+v6>{qnQ=<`*!22qEyJjJegAE7O}2Pzo9W_4;FnLgSL(4f@yhyW}ohWyUAf3*4;qX
zWpcc;(`>=sy~)&jo#1D#)}a0JAyN!J%JOc1hUf2a47a{Waz<}uN-p}C8WMuP<{?!1
zDvVO<L<VHbdTbbd21>s|FmZ(si|oA&KjvPRJ%1I0dw+tVmmuZ_VRNvd(2tu<3!*Iz
znar~zg)(&-aQ9R$#mtW-EG^+SJ~?d7WOsz^^QVdgSN^rDFT;zy5=D`-Iy<xsozEY@
z-sdq?bY&J*UUuNy28VL`6u^JQT2Pw96_U(RW@EoRg!lR=no@gRX1q^<<dgKdenBY7
z%dLmzD0hrmUxJmdz9U8SDk_u9+2{GGlFh+G;9a_iirT(%&xkE#68Dhyb_=CRN;0IB
z-$R7R+;~@;;dn(9ImON*kIh=LFR$j(&))CZ9JO6gELX^>crT%|Wuv)9_CV<MbVGS#
zC~6csSk&(G_^nt36RT4A74?MU%qm=4*;$GedF+OPKGI{<u!bvuYJ*wigI?nQo=NxT
z)bqgusu5l1%a&XCz;bgoocf-DcSr%OgEQHWnI=f<w}#Gki;(>{emBZ5)MKq+)dtzU
zgZ-jI7!-4aC1j64Q~ys`Bp+uKX$V+N|Bl#S`!Q2J0Sfs(l7^k{>DVf<C#ZVF@1?JU
znOh0pxjhkwed{23uP3=SXb@|T_2wgOrBpf~ot6rIWkXmX`J4sy9m@Gt)q(Ujx)Q5O
z7a}My26L|$%I@D-%UyT5V|;r&JioWn$}QJu%4{z-P!@p6UxU~P=T~gfu`8IR^+_^c
z-B$eD4Y>J;aHVoB=H2eXk9@VmNb$V?%Iwd-R2xuoyDIb0E@3k-58<1YGVt<v1uwiA
z4ILLX@;h^j7L_U>zpK2|qRI{XKd(U1;SK0k9YoD%e`MSAzKU4^ReZ<ZM^G)6%b_0f
zIr+vJ)R8_Lia{I|`CX*@4k<_{E<eHi%slB&O?OuH?*se$<1}6Gryx}~ctwiSix6QX
z+$k4AncBSlT;a8^=nz>`cRXjlD_6nr-C4T4S`!;5D8iw~T^3~*hZj!=Qt7QcoNE!i
zsm_ZLrf`J+)A>RVKW5N(GL*<!hTpn!vOsYC_23=$iA7G1Bafv0h>bXb?cNp?llqj7
zOlabTyShntR2jl$k*dV{>JV6`1hT~bg2|y5CwZ(9$yzp~%3>M|u<mLx9e6gD6uTcm
z?_q;Ubz}*3emD^7_nYX$cX#wn3uf}te3TWP!^nO6Xz;16tmm%;_G-ZtOdrt1f2{k;
zW6!+iWl2Ri-(DhHv1bR4-X28x<L1b;H@o3oy*4wi^o5PqGK%OM39AranzC7w)|)Fz
zAO9MH^+yVEF3*fB-BG~LS2{FWr;SeZjicWU$~h_%t6A5xPvGTgL+WdaMK9*5B+)k+
zkv4Lu`IdzLL|&rHB$4Ixs6vCDHR>$}WIL>{mGyff9*T^ODE0X(E9jAp?e?cdKXMdd
z!DZ2ZIEcPwt`u}X1$hr+5$^p=vNZTOrq=XCW<=NQMDGZ^n&^!;Dka#NHq2;bV+b}3
z*p1k=ZhY#QayGiBilk@iLgChoKtie~PdZqECyV3IHq=Hk-z<f_5d2i<p8uohyyJ3w
z+b~`#LYYZK$O@s5QO|We*<__(NQjJVvXj-`+KZHSNm5Crp6hH%C8KR7D`Y2|yx04G
zeLjunzVGusujBZBhrnQd1b?+irQqe<8a`{p9oV`TAtG3T=GB#9X!n8Czugw4lX|jm
z8$0-cwQuS6zpsJ^(Ta^p{^*{5n*0-7ja*5BKet?vHEs&x#@nN^_%pNWZw<E>{a|*{
zfUe&zU<;CGLEWN&?1t+zlk+hW_NWIpeO`ka*+=}hY&0!ZjlyGxe3VIg7UZvx6B*yD
zG-7=UFR~fV%g3IF!rfbVAoiGU>%);gx<jTf%cs{DT~Jj&mezfKD)F!X!{&W-#J>w6
zRJOnZ6NZB-lLdGF^Aj3y(+VoRpTnnK?0K#4qhPBmboYN2-qv7lU*X4d8eO62;xD^w
zH-vinPNi#F&v@89`GUc(#C7XrLwZr}lB<7y@t1QG$m*>(CMmknr&3o|{VD?GZ}0Q-
z5jAL>e2&aES2AxkUz`&AtJjAE5mDU2^)`i*;)f9SW9kKTa+!o<`>)aq&y6yl3c=5J
zmCxVQrIxq7e!~i-Z)9;XsZh2Pxtzn*xU};k9vT<1$dV^qCoCO@?t0>MOqy(d(^D*&
z^8{&URi#P}DHLg4Opbd8Q`YDo6z$PT`X@bw##|BIyJ<?iH9tzy<5w=rVoCI^(?np{
zHL`I&0K>js@YOlVZv44Jp6dIsELcXzdhf%x6K~-Eww-?OPv=`|d$Dg#h0Jo}0C4w}
z?AVVzm@|9~G<09${rlmFUvE!kvAz_n=gTYx-{s1e<tRuk8MW~?G&#8^+28i%!~PZW
zNu~~v_xg-MMuO{YX-DU_4y7u2EB<_LBn_-eK+?x3q@D<)g$0b&wkSfmB8^%y7GalF
znCNn=(@Iv$dnMK9yKJdu2gb-1gc}&k?kareA;&AIe3w6d=%<Ny<K<8au;RJ1OsKn6
zFl1M|V24HpY$xpFsl5y^$jp<@6ud`WVrQN-FoBM)*@U6ZCrI<OHl4hlPM1qc**YZ+
zgo$hEXpa<rV|^T69E)VL4|bO7`<CO$NkbZQ*nsymeuo&F@nT-?2EA{WQMc9;XCJ@B
zk28NIO>2ZdZq6C1nQKblvS$ctZ3jCMzlPWT5a-?gu^4l_1R6?}ShxE(cCPf2#P?iA
zI@XVo+-i%-0jsDjY8r0;^=411Oa#~evt)zbJK61#2`t_0n9)GN4)~-tnxE_M3)90#
zkybN}`flpO4Db13vuXjEdzV9TRv|9NckpNT?+G4ddj227C2+H<0?*yT1KkeFUKmZ1
z?Vo6jQv<sa3Y;nB>uPKaABaJxqG<S4Ct2*G&Eowpf%bH5poqI0kTI%@boAL^W?R)4
zcTYHwYQH&%&er7~E1HZ}MVpEK$93$QIE2^7Y0xWKF*h-rMP_e8MDEmwEOZC+)QV85
zsxO7%x)k~`U@dazc9C957=g`3zmPp67;lX?@<V<qI2Ym$vmQ4j%JY8X+j%R@TzZZo
zC$`evyTWBq`v|W~LP)vZld;i#vEhA_@DAUB$LlgUh-)wDnKrzNjESxAC&l+)B(1Br
zLsw51>qX}Gq}&Ufy%<5Ik`s~#1FrCK!~PHr+K7QVn`x_dD7$e$aHZQq_@KXP<lQ}s
zB*z6GST%|4&fG$r=5VOi9fYNBBaa(-ovf0gkh8XqjQ`eg*&78Wx9A2pY1|`wwqX{O
zkIljK?|Jz8`8!|!@e4QHGF4JEawba7I#9~s3jW_NHR^v%xKlQ!Gb>NgKe=)fm-a}g
z)8~)a@h*oI9{VoNX0Fi-?Xx`7%Y~+Cjil{+Y)D6VWrr?=m}5p`uR<U8%}IkkZmB_X
za<U{-bmkuXt(V0te<YLN?L~oqR!FpmJ;%EOFQ_-1;E(!!;2VCN6O4o7^u1hBD)+D)
z(Sy&zOKUHg>ICz49Xss*af*y>3!#@0!d#X+vtI|^P`Y9)<eO^*XRaUJ{63Md5YAMG
zzi+w9gN0bLvxR^E>`W#JFHt8qfrmI)Q%-viOkLCfCyfh&1AQ2s{w6Wasq18CycF2i
zixT=~eStf<6iGt=Y0>?cQ&`*2YW`$Q610o{@VR-z>F78+q#A7{!yp^{SbCMrFHb>0
z_Z$4>)g=_$JeF<zwGJkW1mig(R5Jc;Jj_3;Kxv~5OpW7Z!*^tJ)0dUP+c+G7t!<bT
zv>*4*Hpv{cgfGjn9tS)Vkh%E*&DwNO^vv!;r|6=jCg(mR=F9PUydSy^+X}-|LG(7r
z4{~igX~S6~d^+_Q55g<SwDKzuo=*e#3LJKtPd%m)re0r%nn$;3XxIZb<!LJlGqupu
zBmlC{u{^7+Kdm#)q=^%n5Rj>f@rOFoO*2)RH2WGht2{*2+zf2KV@9!mZz99h2P{U0
zyDwTL&l)Zx@RkCVdWo4iB9eCM4y8HK&ro?kfmEs+B-eZUQ=7LZ;wv?U%kntcoVd)-
zEv}YyUOkZVR?4Aw<~NC6;RZ1GAR^K0k7;$`mWxKvvHWoA-p56zUMeqLVRDj0P+pS0
zr<7fbkb|A}3@oh;!_yDX;9!5=NKa05qR(t2AA><?F4dw@sbIxc*<_fUip_6(NDg)E
zgqD^88&rP|pHGj#@YRQ468R5q+<=Y!Dt@o|$+9(_s$^dM77F%6C)&QP3&n0NCA}3&
z7<hULX6YToh#i~Z7gb4#GDUt?p^~0hJ)z`&i!t(AE#^5m<4}ty4!8)OR@GLN26m$A
z@EY8aY)6mUov>Vz!crCr7rW^~{$ab4beHxO8j-z)r8b+9R&6*`=`QQ#SwYdF!zBN0
z4en-im39{V=_9W^vEN<zr5p2PeU@EE@x))K`g<F58ai0@B763=qn-a5ZG*dh+i0S!
zh0T$Ah`d7w&-rOf8@gXZ&Y?!=v~I!iucC)<62?5%{f6e9zwDbNnSE&MN2M+O;keb3
zM*LpR_qC~!`M(^tSZ2#>WodBTCgTf74x!k6_GCCmL7G3x88%BUqAjL{oqXxR?5^gp
z-Vc>&#qns~wY3RDQ=BB*_8%rVKW8oXOyR%Zopkh<=J%<H<lR1NumxTxB=z^>@qAJ~
z)L+U;FQ>F&!ih5IuLy$1<X<dqh$EZVP5cdB59D3N8Gn>PJWaBTMZQ?Vyj*TcPj*kj
zi6YTO+PIERcG4G&lL}s5^#H#|4npv)Cm6Ur7;?)S`P~R*n$&(u@KN`{>8J|1{&0iy
zu(N{K8^k|FePpw`O@L4Q1J-rvOIgNdalScEmCMFx@kY^~c5?T`zs+a(o+(xM)%!9^
zR;w_N2PM#zC>I0=+u>;6IO=Avg{9$1{BzJ7w(`RXq+12a#@ULzWK9>WIME2703|q;
z?WFkHMYQ<bZ%i<pi;J`Tsk>GT>PwdJau0c_UA>%Cao0J@ZW>5R?!S;_9!P)oji)Z2
z!UaBijjZ8yCpxh2BeL{k5a*i$tA0dg@t63&wq9^mZzVhXL_RI!CJaK}@_UOEpu1!d
z>G+1?(@epJPaY{5Q5cEvx?R{NKMz~(dckpGXX)|T<=EeS2(=Ab#m1fdL>d?FLawfc
z-90IO-tmQ)amNLFA3tF&&#JI!v5bP2q)Tqeg2=eIjZGZoi~2Xwe5TPVe&<s*>R*Ra
z;)GR{5O)?IE#8virST$*ltt#>lnXu@q_9a@VK`P@#5Zl#C)3&0u<4^NatNaH9lwi4
z?T_QP@BHF_M4rrH#WVaJFM74vZkU+ZQ;0yGkfm23i$6UN?;_POIU^iuK|gtfMC6V|
z)=)v@1nttY$Rop>HJZvx9lT?&=|)7u(B>iQMGnFXTSPWHc3?`71(JG<pw0`@p}K50
z?Z8#MeB29Ci#90KwqeGmO>{`#4>#Rz)7V9?;PRk_ls#@?^>;&>w|xbsty_h~?tYk)
zGaZ|}+)>-Jlp+$3vWcEuS)Tr3x^a06R#zOM7Tarhe<_u}ugRyz>;T;PDtduQZd9Ar
z7mYR}@IkWz`M#aF)4<Vu+0-j^zVBJSvfiGl_jSZK>-jjH8^udPTc~)`MBY8Z3giE|
zuojV9Oi;MUS7$7z5i{q&WTqFQY=bFTQSfQI)yg`JEv2iII#GAJPiYB9@$~*7N*z{+
zq&ZVyU*SP@$qUIm(G#wg8}UhXDt|ZCk7YJ|kkvl?$<B6C;F{~7z@_9a>%H|NBo$t=
zC>fWS9$o{pH8JR+o`5EYiG;0N@nez<o6q;Bg<7uo+OH2^@*rCHha)Mh;0ZT0H|Np)
z?a*8mCvvHdXwvg0ucL)b@ue4!dL54f??@!&$HAn*0Uxx}M3?R+{cs+HWV08tmou7}
zuf-=se@I8)HCmXo<^YY>6Xzh-F<3Ba2<qdNX#Yn|nO1ZU>DhTsEG{BIB4^@F&KBv+
z@NzMIP|d)EgZ31ydzbI+vxDmlUye3q(Fa-^!Rjn8NRF#ipiS_Dr+zj>;Ihm7hC)2h
z;6^!P3-HcwEUum@p+w7<62k+*_<ig={`2%<F6IezEJhIrPELhnP>M`xZ8ZPcsf*M!
z^R3Z{$IdvN)GSM@dm`Jfb(+H6o7m>l5)^-N<U_ZgM_>1PNUV;a_n=q!9v=y_L!RWa
zEEXdN3C{G**>p5~3vE-r3l&3oX&-~GbfRb@`ueQF?6ZGllQ)F2z==h88QsX-woalA
zGp`DN=}Vq4shGVH{X^x4kNHE1vbaB<LGk$~D5(<jZc_mKjwn#sz7T9<-+96k7g>tn
z`l=^gMSb7?>`aLrHJsVR)w>@>x|`^ktbL5EwJOwW!x&iWC5vwVBffItL+DI)%U{1Q
zAG@9}p+^yudG>`S7CS$H24-#{jdR<{vP-ojNWPw*Tls-v#!jI1Q(D>3Y42pC24?ab
zQOl@Cbnxe<M&j{fM~rl>k=YCvd`&a4_q*>#N$)q&r=W5==ctJVo)201^G8^*IRO4G
zr;H5p6tU6Biu(O{j~3gPFwS^MM+RPzxVAO(*b(Q@b6OxYL|b!vxAlCW;CrlO2AH7h
zjG|uW_~4Hfu#1)P-Hpd-(h&<%OFPFML{7Z_b)v!7r(l?QCO*_e;gP1dY{btKSln(V
zJ8*Rtzu3GI!$uG*+}I-FaTR=fH)klPU1S{_1*ahYKm2#ag;XBo;?CH3Dx1>6=Zu?6
z$L4pzY;nF~srsI5PCe!yPk18y)ifM)8h~fIv*|~jA9D!(Xe3!CLDkl06s7i+^0tKI
zbZiIrGPpuROD9X>nwq$g=!~9Bp9iUN3PQTtvO>3uH29hm<O@!-@_GrnG?r2S>pMwF
zbO>|0h2v1p1KK;#k$i$X@H5gEAFb43X6XapIgyCmeS!b-dCRAHE~NflrlRwYXC&EI
zDydiBj+L!`7#$Ns;XTLlCx@!2BV{h@7*j(U>&)4t!_8tQ)J8?K99I;X$U!T1(&_nu
zvhjvj(GX(7w2%3~X@xu6HbDj5y9r)ray9z=i4kWY?uZ!ljr4znvC{oB;G#JQNyUR;
zVzLdL4L-3|MP4lAr8|sO+VZo-*O+ezLvz!5W;<n?EO~QbzTUCRtgZhN3UnJu3A+{{
zv~zFK4bDc7EmC@@;tyH;6qGq9vb=q#$y4<MvvJ)Dg{tm|Fz8yatZe~;j5b5X{sv2)
zv0QS^xD0#Ea&dh@Uure!#SOJGsAFa%TcEp@H)mNf<H@ajk$BHDWar4xScf!r$01v;
z6mlUaFj4U$()2y}^_wrC?~ujh`-_gbO$he(c_pjsVNKiLdhzxx!CXgIRXXK<4B!0U
zE;x<J!nMA$X+sn-I@~1LSnLA1LQl$gR6-4V`ayeb5NgG|Ze8br9k1mDpXCz#hUB7Q
zci)2RJB0I8empJh*Nb*sGl2D}+0-86AS;qYV|rsOjg{!qa<%bn?59MQzD0EC%^cV{
z9*V)wiul5`F4%T?6;2K2{B-0FI?~b^rDylh<MoFb?4KJAk#3->Ugz=cdc5dzccUI=
zf0<sE3EGV1*ngdK5OC!UJHC7`R?cgnM`L0n<NdFZZ;?N<xZB9~A8>@##uAB(@R&7e
zC19AnH#?!1o8KaOcJms-AZM3_?<UL8yk(&{mr8|tP%swQ-M}A>!BjNqDTe(9DLUrk
z&T--4nm0x4k36wyP6n1H%tFguM|%AEG1FJ`qgH)ezU+uA<@H#HRS$1q;Ef05e?Jt?
zH{9sZ&hEU=%w_c4?+`hD<CtLV$A8_vMf|hyDXMHj%~5N5ZuK6+p2g$E=<#Iow~iHx
z&Sq@S|7hi^hy1XS1_fS7Me%es44fK@UiUN594tOlDg`(!a%CA0?^Bo7-gNuYD;(&5
zmd!Q1B8e|NgU&fFu=QGt8)D}dX48n4{2#og?{#u3nub?%FSD>tiEvmTSmfe7am~k4
z3~MT<fA@~E^2x?%jr$LQZ{mnfOy{{ukvKUkjMnY-MgN2Vk#F3Aj~6Z2N}nuB3$#Z~
z;z2kJdxlB+J}@6JpQg{dCDAU4gr!G1axIc5e`XUu;oq5hZdK#2b`Irlc5Bd)J>nh`
zmc*5R2sX?%Q@+6~l6r{wbHUwStfg9+qP^7eCS1LRmn&@0y(ou$^Nc0?1-o#<BM)!&
z&cpMeGRk_SFb9ValKRNtt;V?7mIk<eI*-D~-cXI%fTtOwSW}}C4cv2+nQIRc`<qtw
zLGuWi%$$zJl0rJ)&;@2s=EGxMEW29Wip^U?=-;|L()E{@Ceb1ICia8sjA%YO){2aW
zl#!c{36--WFpcRg&Gk#?CA~68#cw%wPBp|CLl0cJzKf^&?ZDFge<hLrS73Q?6wZmW
z2HpHeNPgP~$E+jB>Zd;Mq|t@G+$l!toJmwSsRyi1{YUC2rm;Imis(tgUMQ{!rgHO-
z>~qjx_PN0py^NmFqBql^SYN|W_ZdnXX8e$RJsO8?p@!sr?JPI>QYVWDJx9)&g8%n4
z2Q@woWVgPvw2N0B_FI(lKDG)}ashnLt~D4u^a<rPT2SPl687M;uO!j8lJ+z|#E;?^
z2<xBEzeb9E@HtJGc9@HM_)Of>j>i}EDcm+-KChczMX~R0@U{Ah?A5$wY`5ERWLN*-
zGcu=h+hvQ<TsJ^6R5IPDTXj6G{>TI$EfA_t;}KQ+PINmR>1|9ekcwc_duZ@F9WP9s
zw}R5VY;mGT1}y(NL2hg)3}s$a5f?-n6T&F>>n(2MxKXfwqVUr~12=|6u%}k5neXU1
ztW@nY(v*mmb#dpUHH9V|5iF|oJ5)b-wCwheTFj}tO-9y^!cVk{=A9IuAB)%2@%aUN
z9ybYA*&>7RXAo*rlJTVIEj5LAXVW|G;F;q9jGNw-PW>JuGROm1;(;;PI4+168I7mm
zN!OUc>JwC>@QzJX8bd4RkA_mu1f++EtlYwG(p2vq{OGb<xZOG+Kl8jcoi0@sUhi4_
zhRBxjh&PxT9nPm7w87wxxzy@+P4;GTuEbN*ob?S$fPITQy?$eb_#;0h$0|3Gi&-VM
z_s_(ijA8ip$er4CCz55mDFuxm&EBm)L~co!B)dLDBi=89PrCdUr&%K1>c}VC(93Wx
zUWGpFD?hj1f!qX_XZ_Y_Xw>eJu&^P5o#Djpg$|;mjaPBS{uXn2IVgXmb`Jk7p3l0I
zeMrA&D9w*^6fTiV+-8E<?~RyFkEf_2Q@xlXO8`5ABJ7`Y8l@MN1ZzObw%p&$&uZ?b
zqs5+hWaNy5xGvJ%^M3TucP&q@d<YBGVrIHen^k8DZoP&$@7tychsrX1sG39*|2u+T
zhtKkf(FU~etv(H(lT9`UF4La{uOx+Mi>ccZC0WRvKwLCa1GRa{;&s-*e5n<Ub?j2G
z=c+z_c8*4y!&;`0G>bK6jpU>E*>SD;Iqa?IN1ruqM#2H%;oVRw_{=M@wCFi*INfJ)
zsde1-;yospl*f0sEra^BTCSMBmRXIzPgbv9(m%}_Dj$3U+*VF{czY||ejQ8|5nf{d
z+W@8RGbto(D8=Z<!fpl6XJ<tIx8xB1Ec#;PqCbw+IJh#CJE4-a_vPICX(G2BZY{XA
zU3gfPsjNZs4li|I$hprY*gv!+7uRlR2yEkhA7=2BS+69zqbn%cdLA|p-AxNuoq@-K
zIka)bP57?qN8MH{6bMrQ`N!1B)TZ9Tj|poq!|guK<;So~zsuOEFN&;}vj(5C#S(J;
z^RM;X>%wO_dEoiNVs`7I*hBqyn5SQwi7xV{xEXL86#J5Pj1s=6adnbQ_sTK+tME}R
zl#rdt0!qCd4n^l<)T&lUy-ItdqBIbG3L#ka$&KE;c4HgX#ef=m$%d^Ej&kK0?5e?8
zSdaEYY>)fQ|KJNOerU;u+Emc*-C?pq@vhpc=Z(Q)2HHAy8y{Kj!`n+=;KX!o1lh?K
z)Fdv)|Fe6E7Ri)pUJ7-$%lM+V47vk_3otAMeFta5e1-+a`pa<Vv&h^g{X+8x`GU>Y
zWl+|ap-Pg@6&~H;y;&F{E}r8lVQ#GH!zoO-*~XPdT9ZT8dK`Ta3Z@;6<Hoz-lC+!8
zUFpi3LwnM!#UHTgkskS8E|C=vjl!VkbI9iX3fgATjKG7{G`!*khHUAIa`}_|_i<Ia
z;Sk8PHg=Ka7N@YS3wuZ{Zz|A|;&}9NF2J7XTo!K-E$O4QpXV>Ar-})AnC|36xto7U
z5=+`J_o^49Epk#z>#>rp7s6@Tt2%fuX=4vmLnY_)WAiUt70MF4eqrjFEp(t`4*6-#
z61|}uR84foVBy8S5fjKB{9BF6%bKLo_A|e2Ll7R4J1)=dByD|S4!Z?9bi7UUK=;;5
zx>qSl8xM<YLqen2Y3zroMPDQ&1wwiDS=yiD%ksD}I=xSj87F1)_Uc3$dVeXl2Fe8I
zx<|p<KX=8BO61b|d9c@|jgs2w12J&sCfqH_;j+9wvLi=JS#F^R6($^!wK(5_*RgPB
z*l#i^ZStewwGtGLD3wgqFUQG<ZMawJAlPUQ$n;%7W_PyWz2JRzeR%=vZ=_%^9B9O)
zU&3V?LD|bzQf8esrK$d60h?_pHa&sa4L(eY?Hbfrpd>x1{gtjNCqe$RH_Q@8Q9tXG
zI4ma|C7pNj?aw1%H}xk1+JYc=VMl)4@ed%4n`}?xLNTjl<F<SRyuR16%XXnW*TRFD
zOU|-RI@S`)A!qS)WH`N@S5GS*#GuL4mCW9XGtAT9kZSdu6-;a)o1bHF{-y_3zMqaG
zQNMZlI6u-EmP6|*w$rAPa8&=yf~CuOtQpr+I;e04yr&yrM6*9G{go@I)eFY0orZMo
za42ul8Y$`j>kuumb)%c%pSWVK5pA_8ki_gY!psE|n9GDhq@DI)DFZ|D<@PMhpVhAb
ztAo~Kz!+mpmiNNEwlGAy`O@_l7W|0He&!NZM`81m5TBz`Fl?kT#&(=UrxVf4Z%hsv
z$3K9Ld=9jiX;atI20miuc3#z414EV5xWQ&4NY*_^cHJhK*Lo*J4J)PN@*TXe^%j{w
zPotgJ+9gFPr*WnE2Il>1;h{SnnZvXiNEZgu#MGsj^lcnI9Q?%10`q8ysXN984xw!S
zSM>kg&9pa-G6RQfX8z5RoC>cXzQU1L?(t&H<2>`n3dTsHx;L%1*@vOxx}bZu_^fjl
z>y?(W1zR$4zge!p$I2NFS4D^aMULc#rSNFXucK3sqY*H>1iLVY<;+$^-{cSM`SC>B
z7Icud4*tUQw!9Vm%0_gU520NbI@7#!+DMKmWNv|4^h3EXelPaJiQNw|eO(3W;uNIe
z3eFg4*^@%AIEwe1A#^VAuH?|>p3+$z)9K~7vHW=HM}DSnC7o1@rkvMLS?dTfw+Ob-
z)$qyaZK98TYo}wk`aXK3{t)d4ukbv%x9Ih)f~WYehEu~VeoU_#d3|Ex6+D+Rh5j`w
zE)KCY8RpwWpEN5SCd%P(iq*%wxrVH+N)C5kPK3ggIiwNsQ5NfV21z9!VA`+?Wm8q9
zP0o50_<4r#VTWVS=2mEa7=sUr!?0knBRSY?ph*)dp_X2ZEn{|ahw5wma%(7#WOzbn
z;t^=yIt3N89<s*)rv<y&f<Mb|LXun`T5$6ZJADjn_d9FZ;OZx6*XYM|y$><R<pXJn
z$U6n^+=lZ{!)e2~aO~>29=?0**o!VZ`2g24{xE74m335*i~nxfu^-2fGvOU-mu;oG
z*;3YRgONBFkx}Kz`8epGDzd?0{6*SpYHK#4$zQsWVv&oa;oS)w_@v9U{ai8CA{x#U
zmr_URb?Q^L2>a#2u&LW2xLbVS#+|>?lQE;|$#-LVp*NFKvI6N}Nen9$o%tDMr^POR
zHxk>kc$(aJxM<1HY2Oume4fO*C-3H|4sRtIX8H8a;1T|)-Dm#<XZL<lrexOl71UF%
z3xfL3Am>?~3m$)Mpi}NHG&y7hOs&VGD)Tdi^p%j>p*n1dbK@`R8GXZUgv`jH7l*R&
z%qR+Obe>W?M=&apCzUVfuz8lPII|X?v#X`~UE-A3MJ0Ku-k`o*GddJESD(T5uPW3&
zI1QR5vE)9;l<$r3hI%g*YMpVO*!K%`q|ySiT+w~$Uk|xS{=6YQN#r6fLZ$37f_)!j
z+yXy3Fu0VRJ{XMCx4w|Jx(*GDPG=g1&seC%D0WEfBwlZ@MnKb5e7bg#l%?*Jsq>Nc
z#rflOcj0fHH%vHIW<YUqJQ><drIbNaaQLbP`|~)H+!vl=%9hvAYA_1T3kS1NbsqF)
zV>5piR!;j2A}H-~F7x_xi2hB?=M|=E(pd}7z~rJX>{2$6wZ~Rwl{<<q?{lJ|oAudR
zt7eie5Db*odiG0x7?l}$lUn#C?wWFirjEUYO%v0gbbmS0j{fE<Ym2!<jSj4J#crR5
z@!tmuFmJ6F^Qr!i3$ZQgM-O9fLV5`=Z6FO;HHv?F5Cyrg0-SH%%;)@*$l6EfLv?W}
zE7m(I{6AhOuUJPRmR{Jl(vP-08q4y<d+Mh5QTPPmf9`cnxT@MvAofD%-6rtJiIKSC
zxdXG+Gq{UoH<$><t75<{HqpeEdRfU!M_;~3CTa)xeXH|yTBn8vPaK5Kn=>%|tRY1u
z<<ijcOIcf|bhzG9VpP2yBW7k|&<k&}Q6EivChZ`-UI|9ge+NtKUBhJ~5BqZ4KY6sh
z!=I1Lx`p$fzwo#bN?0|jU9u)NN)qPti`lmv=T}~qqfWttl1^Qrwco`5b%=0F`<HUN
zpfbo$bwqYTCM{^4KqnsMu@!H;;a7PDOY-ZW+;#z#<2~5fo&(A7|L@{!&Xf1L1g3s?
z96R^yGcUgTjdvFBnSW@!;NnCf=70tjoNSlaY<fx;ZWyrMO3}DhkcKa7R^rr(t@tUj
zj4uL?AwhIvzMQIK;U$?k)7gOQTySI0o))0TzFV?2>aJ+)IFIu_y(sK&CT88eg2u8;
zvTl`7)U>~BS#DR{PANuy^e6U|ug8{*K(@TET*1kbTQuEp46Q2Z&F9bVME>^u^XE0%
zz;~PyTtfXRWbQGX%haK>l|BfT4-!m|aacMbo>%3~CdGcSe5i9e$&P(QcKSdTcWNru
zR>@2KTR8gb8{ieb(w)LEnl1WUaw&nisa9U>utgG!)~e<KX=$W5&5acGZy}&q?0R2W
zOMDIwWU7Vs*w3z^@L<dI4|+9{ep!FnboXv}9ybH(-74@W>K!?H8zNN_#E$D{qKo<{
z#O(RUW4{=Z?Tiyfi2uZUWtngVIieMoPf1drhi6t<6#DKN8C$(U&ojf}5PJ`e1`pWA
zTUC-VJNlFTn&sHOpb+;(uRil)8#la}L=B5Va80F^COBOdDZO2E?{E*<0S!)8ib^zD
zKbt#Gn<)B(MbjrLUE>3WdNET?C8_pZd$L^X&2;koaroqA*rz!$57D(>`DG8EG(8i)
zME?JZaIKD+@SeV(OsBII&hWc*j}18AN#wztY5W0ax@xI`z0ST!fBTJVTQrj1sN=MF
zus;9YX(gq)3$DpBH4NAl1+^Ai-jr_4KFOcuUAH|KoS6%BvVRko>;H<&J!-*8)mK#a
z>Vo8l*ALjrBCzX_5zMy7aW%nvoG+LXm5a}y$@nc&Ek|I(ta~te{2m1p=g}0~a5CH2
zD$&@X1r@<gj3}6ZTICn4zq~6>1VpgEV}f9Dc%7_#OA}JAD@wHz9a&cOD^~S)1g07I
zQ2T(v<T199bh;+OROK8@vy=GKDeKX1{R{SWoj>zG8$qwHi9X=3-c*%658nh^=+Ven
zdO2no?xmG+bMq?fO$(#$W;qyEt_syTS8=Vb8giR_>DR#=RG2-W35Cvrzf#82x1Poa
z_m@;W@G6<8&qO!vU)0iZiS!b)ap=VrythA(eqBf4-WPd#e>;Rm+`h_##v5a&<4@)w
z&8LgUx3k!J3R3Uyjlz#K4hvE(QCK_xV}Hnz_f+A?r8JfqsVsGS7e-yf^+~QVO%kW|
zk(>T&rNG82STFac*IQmvL9eY6Z+i!{NS@(Yu;^Oca;J{R!{D`VGl~{=qSUcYvU4-U
z_obF%;&L_i_fQL~8?cVtH9g4bd=Y#KoB4wzHTt|}B*q*qf_(T=@_VOP&_`2^8V-&`
zgU>an8R|*a-t?qFyM{5ZUkB*f#2s|-!6}Ln&#BRadP$$H6Z7u7-*^>#4v%b<3z}{F
z0-=6lZ|5i}$cUqb2@Z%ks#=hCa}=!aitbcl9^3Ql9;}Y1^H;kosqN5pTJSRs6$=mZ
zT;)(BxuX-PQe-<8Tx{U-7qe(o@@)9;dkW2wmocmEB~o&B(u#5|wtRu2wD2A?db2JC
zVeQN4@D~WqZ53<&B%F{ZyOE#5NYW4;j(H=L3l6hR{HL`O{pT?rml~RBSmZx;ZI8Uv
zsdFcBFYS%*$=CVjIBh<!_b0ZV?701>cJ603iB{j4EGyih#*J@zK_x(NXo@n~-LsaG
z<`eJmX{aJT#1=Ahk*9j~P@aypz2lZ+F0h_n3exYv;Rx=j13g<`?lkHK?H{*HGDk%w
zTQy@c?;U-RHobMAk|&#Zq_HVIIyj2yuI(lBSa{56U)*@ndAY;~Wp0IUoFPBQ&PqlO
zi^U9)CwZ`B1|qLFvu>|y5P7wh|FMhaSDqQurE)DEP_4%-CZ?m?)XB`aXCeJCPe-kH
z9_wN`7uT_$F5X{6gGM^aOc&3Wyv_7xo3>oT_^_L3yPL{Qbt74P+Xtgj4{fN;d_POG
z%tpt=E<E+R8nO=Rva>bT@Y?-^#f4kItNjHuMZa?KwpVx*nn#X-XDPq%F6K#ZV8V~*
zOhx4k4Ve^$vJ4G)czmY2;yHhMnIFGztqAo~!}0UpGCDSNFs-b7Px+(bc=v%R^lZ>c
zL|rRnQ@Xmr>c9nZIuwE~xpU!h$QuK*B4GUZD|S75$lC&&aehkzTKhEPcG+<5^FU<e
zPDP=vYdfp!-N=UK>_+8NOFGzR3`@w#=e`GHS>EwKJoE4+9E;9izkJH+l2o`qe|?dC
z=@-mob~o8*J<)}87)?Q5rx4-FIlo!J2Uu>QFJo?Dt6mecU0FhJ_I5)-%NN#GzlaCi
z=nIdb4=Fl20FATUu_0mw?&{dmb229dRSoRuDS9m|oISC6Me3InqzlF8zve?dulq7f
zl2LUY#&e$L&neZ#GYca;++c@|ECad<-$?88D{^zMA(zS7xFPy9Dpo17m2q9j{8s{*
zgsp|-w<~RM(q!@5ijn(93%P6ll6PGb9o)7b?WH3~Y2ZrKXqj>=!^NbS7Xytef`xo^
z56UAgNKhWgAT5J7tqP~!BlEDgBY*}db}k6yZ!uW+C!Sr{BA$Vrq#s&ZQ19#tl%C>`
zJ}g5`$Po6_axEVkm%v{rHluZ^9rixSpe=1)l-+WMjvn10@+6tO=b?HmbgZYzp|d4s
z!o4SFq=(P@G~seNpZBk5MBmdR*zqN2L^nPQ7o1Oud+Qzg=^cke@g6odyO!#;+?e$B
z6D+kB%$dE*+3X9m@$Qr0?GC!cOLyO3Kd;JRgGwZJhX@~xni2w?57Ef6R%jZoPmW((
ziB)&Thp~j*Zhx`ET8Z@=s3<M7tAJkeU8MEfh&hYDN^X61rb97_yzBfHNoweRO6jSA
z7g5_Vxtks}#4kht&z=;z<O5%ICXcNROUA^FnmD;w@VqQ)=%ev!diOSi`!^qvZAlo8
zVje21JvIz0wCfT5HN@zRuML}9dK)1>cTiT8udF>f85>h<;r8_{UU%__W_+Gxs#_-V
z!x!_Uf;aq*Rv8~_pDH@#QZeIcWA615Oe^^d`-^9IXvBQj%Ae+yF|Q>1)<*H%d^5hX
z*_$SNT!QKv;TW+azH-t~42pY-n?su6m-vlWnyX{CRRDX|u@~KTRztC42b+Gal1i;T
z$kn#1^zjYR)!4@|Vr4SF5gmiWnw<*%zVX8A9V!?%q%&U;njmt#bI5DoOWsyJn%2%g
z2)XcVSZfyI1-HkBjJ2%e{t)I~9fj=jP?Cn-l;k|~p&Cya-&at>3X^ZJ^lyIoHs3tN
zIfQs#p4rVDzIT!Cl7Gg}-*m>Qo6$6U)e_7~j)hxq9f_QC0e|s5ME3OcJU&(I&By&S
zW;>k48E?xn#1AN9$wR`~(Ms`b9c6{Te_ZMQ(j(NNae<xe5s3Jb8Nx3-gH~&wqom40
zyoy=Nr+>ds-zQWsA1zOIZC^T@H0>+S&i@ZXr*@Inc@2i=)P*#p%9pAxSmAx*GTP!I
zJQbat@N-@ozTc=I6=!W&DmSr@(=74oYbPkZyF|mwa>d`<3z^4<Fq6PV=rK^6Vm<D&
z5!;^1HirPIvwBGfIq!oRO3297oju4nfGr<y(PO<w$T{{!<{vOlFtCHEuU8}Xm?%Pj
z&lf&tqTp(T9X9ec8$~lU-!X6bL|&Q{Nc+yp)8C&%=)reItaqISm%ZJ1#{x^bxhfYU
z<-_pCD~3F$I-qvgb4iw7gCxhf0h`RESYkK{V;XPr*H<?3%|Z9^VAXRv-sdY<wJyP`
zQ!+Ge3xM01LVi)@7IrX8yqhU{8BSf9sgfOQmpCBtnHMYRZqAz?<>Sb}d2Dc#xR=hp
zEtp$hp?hsS0(|78o1b_IAM$Z*IJTDWU$zEg7YPT=CN~!FA&L7e`%O!O58%q(i#Vtn
zj|SZg@~L#C3(miIzQ;P~)>v?bjjGbZi)SRJcU*YWDi26r#$f8=1n4a3DlJTdM9FX*
z)J>-{Lzfzvym0Adk1^-foiuST)s7wc9LKb4+^MoS6?fMRVGeEy*V0e^6`7;oENb5~
z$W7i(W``!REX`f8Jf}{Tk`KJMbUr=)uQMEySK?j245}ziVRsg2V(`5_$k4ckC5^Fg
z=oZ7uH5a2(%r|~|MmakDi<PMadQop%<$@0lLCn6+8Y`Q@N@|-iz&4)l^z)`^Gt!V*
zIGt2Fi%#n0>%2nAi|_8!71^1o__?M+b~|j8;F&$g(Ca4H7$WwWpH|WHewi4$$ruj5
zH4TFrrt?1DuIPSGxM_FZrjj0GNl`^Goqq42GiwVdUOyfkp#@a=Upx)i-yJ<>1d)|V
zXEOe8W!?rg2mUmEG5MV`qO6%ZTuosD8NBzybdf7xT%W=vi+-`-Ct^;yGyw@khM3-S
zWWH>_F&xLGQ@8P;E#;?iEFg+}ZXHGY%OM!{BTZ)W)`?e_HDjGx4vn}&eDbwk>_zoP
zvd`QD1@D0@;l64?aefltDD8$QJ8e=t7LQYY_DD@F;-@78(JqVR7MUUV>nomM6ec)w
zpZN0G-C1ehkr=f#hAYpLQPeSaL>(FocafFVUXjgjCn{nEm*<6x{mF5~I~rS{AseQ%
ziu5EY2nf*U9iwBh^x!P+78U@@!g%U#dtZ1hPGh=NF|Lhv#KY%NGCQ(Gd+;~>aNk3%
zoqMA9C@mWDSDVJn3q+vfUfk3ZS%{oWihMr|ndTzbrm8^3B+kRP24kUp6LP=&;l34`
z?C`5pY>SqsyEduR{&RKy_*Zu@)kAQk%6v$7h6M6zhp?(*0LB-8<VM$hX>ai;)L(AG
zrhmOC=|u`990`MDODx8z$6#Z69=q`UCM5>P&~c?_=qkBOy%ZHlWrdWcpD?4K#2a+2
z$rZB;p7NN64pci|L}zI-wZ{eE-0pB$W?UdWn6(6TUEZ;nQcqqmHG<ZUh-PmyV+4QZ
zj>w>=^X_UlN#Wmj*78M%e71j;sIU9T+LwmGbLcrNw+@m_<DI0}w%kQw`vJ+O=5u_h
z{eAA}y&tFh7Sg*<d33UKAHh8FLF1U|IOVpSf9o1go1Ev6!xBxlZM|SlimYT8?`1F!
z@(`T6lQ^z(R(N3lqPRx|W%injgsK=Cr|wE!#2o%l%&0oL_t}ZBF8J3ygm74msT}v?
z@yMmNZha`(eHvz7djVY^!4@>QOCBzdFi7Jsb1j=mmYqw<E%%!w_DnZAI9Qo%pW2}B
zaUVFy1<}i57r{*4A>1*c$Qv?%U)cMe^*HhkokL#Blzn4xV6!qhy;?~3v{SHl^kx1@
zp#n)scD&EL(=ZWvvem0j;owy>uKMvA{=681=%t*RuYZSm^#WGO2b0T6d#=);gB|)s
z*t6*hnkFxXXK&D%oR_R!egK^AykwnyzcJloFL?KL+sNcoKJ9!e7^m(wIF_v-eJ-x0
zHMKv`TC$nHv)ag9N?+0Pp_K^kbpY?juO>62a{6;vSz06T8@(cklGEf+ce)hA*PW0Z
zUU7kn+Qb>tqf3(Q7Y-n5Q8Gp!=_c5@3(2?BBE)^QfvjdG`j2s-=G}qR!`uXq=jLL;
z>IXb`gbv*vWlRI+`y*>#ndFG73iYoK=E`r~sb<i4u9esc|Hl629eMZpiMkDNuSkMP
zR;XlD|373?^&d8z4nuI|0-3x~HH#0(<<mtjeL-ynE7^Mi9`K^pAKQ?ZVTIr4yim2O
z9N~&<prD?M_%8ZvVND{Y-gu143$6$&2&Dmeo7m!MP8e}eo-g?wk6UFUaCLPf*Tqwk
zAH58hB~syzafL;1P1aVT4hNV0{LrE*cCUFSb#}UoDFgp8mGyzLonzl&|8gCAoN@@&
z%9q##IpN#d?+v=WmzFHn#0uIV8}2HY4-0fz*GA!}%NoqXENh{hGKiKg^1|Pk5v1$l
z$@bj}AqVdZdCEx=q=<X^ZQZG|PMtrq>wE8VtpF$LK7THrWqhQj-na4N#A8_f4u_Ae
zg6w43P{>aZzJm!~q_z4zwL1LfX3d2({OE2(dakC8R;%deSQBJwO(o4M{wzZ+64@1r
z?5%Jyq}_1iSsAHtU$%{_7Wbfzb+72d$NLiFFZ&JW?aD{_C&AYn_X56=g;bO80_~P6
zTC&xX_i%85O+Y%lB0ACLv(1$BL>ZG>A2B%U@FVYmCo#cLe!l=785IPp%UCtHns)y(
zrB&gvEKG33mYldpnqF^thuj(%Kd50^@A8pvFb;c8UYG5iCpHEtk+@lIi{?JTyrr(u
zXvGT?3KKI-(X28A&pQH#2YxiwVjL~KUB%w3aUK%3pE=7<q2DLwL08O*VQaP^bg?O3
zPk$t7l8J1S`cAk!bHMeDd-&EnDfrivL&Fb!lAOEs0=*}FWRohp;>GqP91QFamm$Hp
zcgvhd?v4VTc47|77ir47$Mm)(NLG0=3dx&wS(~&1*~dPzlcO@QZJG~ymW`(KwdUB>
zV2+F#zPMV~6G?sbg>yX?t3|iGpPjPwPkv8n=<l~|^5Gyde>Dz@BMUIoR!O>5oK1ch
z-Un+dZ^CFmHd<GiQ*eM2GUo@;CB6B0T3QO(u%U=E&y$3YT7_S_-3!!r#~@;wDgDc|
zM^iy}@hk|H>FxfS-+Av*G=FhqQn?zkyXQu$Rbtrd|BN8_@r+?-``!gw$Bej@<rTKr
z-<3u7*~Ev&jFY)N6#i6=V$#SS4d+Yh6nQ-q?KAvn*RoCpm0SC<EX9k=Gdh>%9`}To
zRiL;w7s&qJZ^!)2BWdoi1+cak*GT(C81^}Xi-tEvpIh8R6cZ3UB^t*JN7BI8rgTod
z8nYj2lDF8a$g-B8!0`Z9?$$wh&L8MRioHW-IV<$sO`4|^rK{VV{(sg(<=6WgEjW6H
zhEH`N)4EhP@q8^R_r-~ROK;41co!K@7vNS-5(W3vVv?1vf?IEZdr+n+_6GR5@DR^d
zh^MKC_mk(ZCcfZYX9_tUNA`w^tkkYV=5sWWyjHHpl|AD7wlAmUZxstV75#)oneYVs
zyb1x|r<`w>X_x(a`fPNPbR0kAXP*>5uSPf)<#yyZM6E;PKn=Q~CLS4XJ!trVFnaec
zmu=r$AUhCulMi|?X02i=Zr+oLT~-)u4!WQ;w^H_J`8-*cgCp$ap5a!*SLWPbm8O{7
zL8Zz&zO6Y0Nw2x!>Leq=)R5Y*970B0ALyRS;-el*Ve<G3HXgmohi!YqLn3^b#w8V$
zes{sn@@{yM_7Vebcrvw#`LfzVZRnXjWV!)f7^*svhsIt<^v3=)?7k|#wRy4D^l;3W
z(;#ygVhAS#(Rq|Tl}OGc@jtWdBx`>?!Ds6@7Ui}L1*N&-tXPWMneE6;xK4@lO4zlY
zn%p?+r;-1Z!Lq^KA~B|_fPVI<XE{se!6ea#5)ag4M|?EYW=$u4ZV%qByUx~G<%ztn
zcfR6(&a&L9WSCy>pptvFkbSG5oi5u@roU3M+B8ngQ^OHzo`&jez33hfVqPKb-00l{
z3jdMH-<>Ezw-qgHRBkNx`kJF@eG^Z4VZltUR6?$EJ%2w!IJlqZlK1g}%yfPmAFJTR
zAKb{moNo>kBF>{z)f^?tV>+X)lMOF3ImnWIKsG~DNVDZO+IK5UHy!5K^yNC_rnd5f
zZ-i&+y9K$mU10GS8{jd19JX9|0M}Fn>59c-rU(f@&F=3k-y)XPG_=vMkqv~nH*9Y2
z-_T3k4atTGerr_)g+DvbUYw93D^K*Iv_$Tt_9we`!;ifEh0ox)LP6_MAKcMYz;ZLu
z`MjKs<9F{c(+}S<`Hn05QPK_9Tf$-fAei-gX$>ExVz#`vo@U)E<hS!f>CQ!wO__ch
z(esm$E#}}ky-)G0-F0Z<PI=P%x*Iq9mr7y}D!~4FF1z&C8!aBfOZDgj^mNwJ;plEy
zZ+r>eZTBM9u$<<M--z#;=HwB3jnePFqNOKZV!{e-`rPvZDSpj^=Bg_+>eN;KYQsL*
z>SvP0%^cpL`HCdHPs7yVmt<Jn0{k~(BHylJLh1IaaC_t|iY*&KJCtuCzjG>{sxo?6
z@`b56R3OwM4LQHk*$3giO9?LJwwvlCTejKb)3#a)UB3ycMPi5VJPxB&MR$4bVvPBa
zL5lw}$?!u@avi>ebQWG9t;=87lp<TI`qGZ;fr9z@avwJIJWmPBg$JuCoECm;$0o6V
z8z>mNzv>Lp|8@ckb@ZS=omXOk{w;p$`3}m|FvsKFv7%4p1jon_cGB-1lum9%rvE6k
zj%>iOuTu8<PJiC#%}{bGil!;QYe{RR8`GR*AiCP=l;{5(efQmgO@|7Kj!woFMQ_xv
zJBdk^CNP@ohRjP<lrOrw4^Ir17$i2pFV#nuy=NrNHduq53jY{;UxEGWWC(k-okDLY
zWAB0*#_VNGt4o8dc6K_x_uorRdc<`0zlNHDgCupQGIhF9!xvqs!m3)4-N;I&Etf9B
zY`8yuZykcsr=IaPt!fI$&SG{wgdZctpLNm=#2d8<tgvGYcVD`LCXDYRT{*Km`gu4Y
zVYGtuY{^Zy9_$JAFQfUt(R$=E##r`nA;-QxMeJU&F|GBg<qfu~1+%UUrPYzdJAHD2
zlDV8TWs*1UuQnrX4`<5y77B+WstDQ?OYc5Tr8$cyl8M=DIC=Nx{nmz3bW9Gm9=pJL
zO*>1fdO5T{w}jF{hLNiN4s3ds$-Q4SVy66SyuSAd16)OCL7|BMtv^HZRm(_O--i#8
z8zknO|KQTAffj8S&xT&##NNZ2Ymd_7pMMwAU%~DzyYUDu>R#v>`xsZek5m5E>p-kI
zKk_^TA05|<>*y!Vth$6g7j7}Nsda3mZ3*9RU_iF3GchjaGSwuN(S<#)>4du;F1-!I
zTBQr3OXvV-U%{h3@o<<NNgA;$$#DKQ*yRY$t@UWAJ|98KhazdVEQ&8!7KEW2qb2Ws
z)F|fLDjGFy59akM7iaOy_)F_>EKY8w5reeR??Nleqhi2+cEb!YBX1tHfjhVlrXA9B
z+41dyMfGJnf4*4w53DY8mEkujH>L+RA515?qQlI|=NFhy5F6||1EWkIvE#+>*sb)f
z%zV>7+S*(}rHdk2n?W|lG`LD4T4!O#DpzcM^O}|~s^$ms<uPZ&TM5m~AtnEnWO}ZY
zPoMsm6*#xCE2>S1A0m2iZ|Bmj7Zb3zz7YDuU2*+;PjqW@qp@oV@;&C#?t$B&(Puq{
z7k^_OIcfQ7ZLYLH7SETD%%WzQdEQ5xJ8ZxD4y?NMg0JkXjO@#46nY{6=ihlDWz2P0
zD2l$`-Wih83F5x(y$7G&&hwzN@jUIpJ+9jxjx1>+^ftxdMeue&?FlP)I*-lv^3r+7
z9q3x23)}Fo8eN`PGMiDyF(>j2?223R6?X}y(8ze$Tc4+X*B(P}Z#Y72Zo_c4Jofay
z&m*fvHfOj4`h<_d=#MgXqh|`tdd0J=MoF@&3jx^vXg=-Hme6{W0{Ucn6_z3oa`*mr
zj5ZZa92+xQ^<N7v3=F5oGpg8(=M8Mt_{Z$|$s_FN%$01zY*Uo=ZbDXfBTSyL3*WmK
za@SxV@<<vWvHYq-J-<G~W7QN=X`jmtT|Y9l@*$W!_#;IqMzCHz-^gyyca|-2GT~MC
z?&ECETA9Y+QCzhx6907e^1-%2w9+7%t{%D}36zdtAseDFEk1_~UDqR^dNy^b4W{)o
zBM~y^J(_f^d3<Rs4gFNie(1mD6RvlWb|~keu4F!TZ1I5VkkzmnS(9IoD?BzS)hzjd
zKliSGPwRf^GdV5c#-08>-{OB1orhn|?;FP3LnND|vNJN1=sedk3K=0QD<PX?m91Tr
zN`s;uA*Ga(I?o+RC81JOl29q@8;WH8p5I>}uXE1l^E~%`UGMi#KhuB_5vcg>z+4|B
z@t52FvH$bYma8^m+*1YVq>?u>6Pw-K{;L*ehmzoxHio(jCs(n&D<8P72<vL|*to6*
zWT08kzHL3jK33+F^ZY!%U)hH)j_^U}kTsGlpH`+lN{{MFCey%Wt?Wch61*?JW*vfy
z)mLGNtmFD-d=9;VdW}%hG%ls!byG+uSMW8DWI|W%1uB1cr4<`1Fk-DCTGL|4TzogQ
zXJ3bLs|&Z^)SbqSROJ6&C(uQ+y)rqsW>)nyil2>gqv8?QAu~KLyV&%DMyGL-jZBvK
zOOBGeuMuT!mSRF|C+N>8rnG@SXhKf|$@M3X_`a0u2t2lc+l^AD+_61rl}{gj;+YBk
zI<JVgTd#4F_lH)S@FL&wlbr0~LeH<xN5|L(KEzg8Qb?=#w4ZOWH6xo|hty%&C*i3y
z6s(etgRs55YU-Mfc^LajNjmn_7#tgRkH`L9OZzvkN7{@=TGywGbnC}^=&t6%*I5=Z
z(+w^NzV!`mCju!eYXm4s#&!IkkmJR>m{FNUZ$*zYeAohXm^bm-C85k`sTDFtM<6Be
zK8B`mLQjV#*@D!Q_@w3ujf!S0brOomf-TIdb3C88CKX4%1tKi`H<Shii|oGyo8Oo7
zo2x3I8!-`v_q))unF+En&9Q93fLl-*n#{I#<Y1<u6V13joFG1zoIy*t;f%?o9OuYS
zZobZ(x)iaf`yYAhxM_&1Sck8oLq2|LD9w8|nzDRv(72oBuw4x_+PK4cK&LFb{6`X)
z3WBWWWmE^=XG)$|ut0MH^!q=NG`^nBzf~mB<Y_Ln)$}Rr=<G)cPer$7lM)rzTtlN*
zFnlfrp<-!Qnk{?JZl3ZbZA$~5rxV3XvR~7r@n<Rf<aNP$5L_U3mi7G4P?nl_2k*ZG
zlS=zOk*_jGROw#sQ27c8`mdlB?T%+Zmr9PPEGO53X*9p=DfRwg!=GKfK;5%;kZiM>
z^wX(5xYVZ*h5N>1%cp&Gqt{Q~_{xqXwu{NAWj?yDe8z4L$V6SuWT-Wz;NyK~GW?u?
zwOvZNr}GAOHmib<xnhdkWpVtt>1ABq{~C$LyJTmL_X%E7160)eQHA?UIL~Y`EqtN^
ze*m8s-vYZI&#VU3k<Ozx$S+GK{dd`LtIfiI*Z;w!=~1?O{&TEe?FNmmPnklfCm&l_
z#5Nv&$9j!!L1gY3sLq)~85csKSu413_g|6QZDTaKjmLbK=Me2FsU<Fw(c&pAS-&&B
z>xaSQ={y*_Uc{}#R*3uYn0!9XqNeiqytCO-(iZu>IXyJdd2|~c=n$Q!hTqt#62$u%
zW{cc_CDj>RrO#~-nC8+Bw%S@S>&73Zcf*yWH&%3`ea8b3Ji50m+bSPN4_45f!@8_|
z+eV(WGaDOk1@H}hZgA5x_T**}OwS&CVUo*&UzQw>xJVCD5INr^+rF^ZL*`;t>i|@L
zT}7)Kevv~)Z^}8W#V>SI&q+yGhmk*RWqwYDDEcxQOY#p2|BEF``+Z}6|DETb8#iNn
zxI8vI$dKt>oeSlN66{IQ1u|~o;^|w6ur!2S4_&5HXN|pMuQERCB$}u@eK}Rb26gHz
z-fLe;(swdl`8kPSDHBe`sIM}ezvZ}k)&MT=RAIEj3SEzW;LbOHnd*e+Qp)Lg?8&V}
z(vN=NOU!6TLn(zBkDzgr-Z4L;LYCB6MR~mf*+~5Yw#w}Qp7**6tvEgG7^IZ5Z?FnY
z(yM{}A1&Is!yEgz_m>$tea5sXJ$kS8iht}S*r*AUxuuI0E)I%Ar~D*(*;+uhHA>Qq
zZc9-AU?rdN`UnrMc7Vs~dRcHoG)7n+Mtk-#JY#3E|LGYNYX=L)(*xvJm-DRO3n=Bq
zc<jG>1odt)JWp=`YWpS9fd>^(PdbYgSN!SfOb@QFdz+3Qf6l`9jKZ15Oprw}Q~TqH
zJ&~KJhhGi1=slQkD^cK?8+ssL?LBF#@24kWiMTt=n4TPy;C<f&$c%kaK5aG5x`^|3
z*JTKqF%kn@9q~T23)e93&J*1k`kfxc^*5HY&)GfkaeO^ujmL`YyAJ95Btzz`Db4$=
z0gX@(JQ@^7bDJK(?xNVq<bK7%XQ`}6xR7RL#WU?3O>~>NhdgVd$lUfi9j>~96Go?K
zNy7--72Vc9m`$HU7qcQUW903Jpm=W)bl>j8sZ<#Zjvi!R6#j9u5gV!4?->1R8I256
zd%oaM8AT5id3?7UxLk6ZjCWg+z}!IVsV)Xhn}JjNy<j2oDRMi7L)^qfvh}1V|1@zI
zy)isa&tB`(9gBPTB<?1~X+hL=zpzsW&!#n1M=|)Cr)>3$Z@4zuiIr=oVnU!7Oa}~Q
zvj#n&JAL;;N%*PU_RYi8KbvXUnNT*vli_r72+pVZu>9TDSZ*DTOz%R;&4Ln)*?b?h
zW6v?CreN$}WJpu3cd&a`E6~$aPTDoW8D)ptB#r?EVu!89ZbqG9-=}#?ZZw9Y@>vWI
z$h^$TBNQm5u$A6t&t$RBPD)e~Dp|Jq4UFjLPRaWg!?1ff1`N3;v78k`&C5Mu_-h8~
z|A^!N+<GHLd7#98_yQU?IZX0?rWL%`h*^!^Whz-OW5FZ6p{|=OD@+wV3N3$Ve%uYK
zt_yhBGbQwDjG(#w>Je>U%GF9Xa9NK#eC^`vA}i-C8Dp~+r*A!D)`9o%CE&c6<E0@l
zM4XWfTVUTm1&6J2VPgC~+hJK24jOrbTdGK39+Jz+KN`%Y8BIjWtnqB<gVStwd^Wa6
z_|oR{<FVYZf?hrpthb|E>A)4@J$=q0)<*a)r$^H+du3^_MOUb6n7Y(GDW5lgeT;5a
z$vpqCm}h>{l5TBiMQYJWQu}2~`jcPC?o`>}(x+ge7vcQw!8P1<WeU%Kw~iH>e1vMr
zU0MBFF?)*J#5!_&(cFDy_-}|4`sDW$S*Q>$9j;1yySc!l^%_09^#^kv7t_9^C@7pC
zz~;|Bi@nD?ljHY~IJ7h!R_a4Bu;DBG4I0R6p*&nPg5a>K7GgyJtMlD>x<U}ma=(Iq
z?|&jg^dHp%rbyfmj>FxmPNX(24<_}q`S(L*(A4b#ueMX@K3*NK`oAHwk;UxzYH=3&
z@Cj<kU#Vu~D&e=g#9SXnVW56IJ67tB3pQ0``rwx6JZG}Lf#;ZR=}D@+^P5fl708>8
z95vN6>d93D^LfJZbJY3m9`>o1A3I>Lj?Co_Y<~M>1P`)gt?mX?7?8r&&s9MAgdSo~
z<H3h4?1s^s0%f`H9BG{KZkks7jfIcx&caku=(yqn6yFY_uhMAhB%V!k&fdfG`}(x$
zzQ|8Wr(lBlElJ*ta<1(^A9p@fFxTTT_$1zEd&10dXA@)&WuH;9yc?9xh_1vGc_^&A
z#0=ZkQt~ZJSdLmmyGLAO|4QQ-6|H5%jMt;kKpWNz6RB5<HERE@gmry3HuU9C_%E7m
z*`J8gZ>MCQmu2j(qO$Z^HKS~ueEfI&1r}$8@-#bnX-Tt!wAwm{P52QeaxltL-r~%^
zttuz`TXQLWmtdM{>+z@iYMJM*4}8PqXGlz0f{yEpaL1*8R!WsST82hue`}Oc$)=^S
zpSq0KKDrP6@Dhp1wf*?p*@=|O!(>%gTt(j83p&~kXxN!8u@%q!2`$UX{?|TP!Q)ft
zqxBGG$)_nIL~wP&wy;RQBXs=dA$sx56vM71!t8_vt{jMg!pl-#J1ZJqC;gCXJ&m@G
z&}A=6Mq}`Yqx@l}3nrKdkNP@Inx7F#31)*)YwEz8$wTmFiYaB*eD+*F7->@#*_=jY
zB-%+>;+i<zjL@PHoBP0d6N8SJY4VITx+Z#YC8}Fw<2z4guEzum<aNC0MeGqBSYuo)
z{K$syh{p=+=e*l0TgYkX-ro93e1aMII6qfKTJs``)u`TM@niLQzf^hn-aaa^VF}Fs
z#8;-gehXfynIUXNI@^^I$}%lZVq9{#?6cMbDwdj{#?}S<)t@0MQ$bqVE_@iHNAeVh
zMJ#w=AV%EEr`opHv|sWGTT%rZ<dF<z;$CkRbc;?O48YebN058-KNdAcbX5oEv9*QE
zd6KzsX_usNM?YuT5}&iQsY@^&+SdyiyFHoZ=}=kcu<p`|9ru`=$y{`lI$-6ei>$OE
zAA<{0u{}5)xAdbh=|~A4MirrD+YP!`xD8f{S4m@*KZ+V-aY(<2<*hNNl|u!?N%Iiz
z6Jd^me{b2Cfl~BWTZuuZI!QZqwt)N6%`|V$E1aF1g_CK%Fi{Ic^y*d^+^Xb5ye#R~
zY<=8q{>3|FhfOUy??7goy42EuA(eNc+*rYVoTMa^FYrT?#XR_aID;)i?$Hjx=g^xw
z75Bt>dQ<EuoNEl@c8?#)w#aQ|;iihx+gXDJTVMm}?1{$Zi?5*2d_{7@MxG>jfpmP>
z4D7r*lH}KVQoHRKlIv^<_w)B<raz7ndWOUE{XZnde4@qP(a4s4l|0b>%WaPKM}4>-
zBx6)j6LFg^9`L4Hd^bkEFXY!vQzf>pMUs8rYh<ZE{^R`M8P;=Mk}SV`3iLL6;=M%_
zbACRIRri@rDs^@UWD$51uv6kOU5BDw>T$hL_zEtK!I6RkRA;Io=J@|nN1Pqo;;~q8
z*<}=RqZBW0&%(2|blFk&P`=VRmpYG(LO9)G@qf!C-rw(0i(#l>%S>W}CaHrfRU<K`
zt2D5Ok8oIzf!czeg7L7N9@UCYwSAH#jyYhk)irK?GK?9eh9mak1d{7kM$d=)Fr&;9
zRN?i2tU3l_^vq=lem$Ch**Fx=Q(i)GhXR`R)=Av6a`63PJg*3kp?)EsY1*qiTstc|
z(RU8>Kj(5`yda0ytPMuJxZ9~r*~D+@`=Uk`Lz0jm)KlRT#*M1R^Y<bn`aTR(vLksv
zm1=fp^R}$iN`3U+WKDfXgl2bUK~(W*4JD>{@maeZS&(~AI`3N}W{9d}voM@~c>m<b
zwm+e?#p`MFnQXL8m?j$`vVZ+&Uci@a?XZ0_2xkHwv2I~sv42}VcV6&w%2u6a7$!JA
zx=U+iBg>Vkc+n^}VMiMEUU-Lgxtu{MocPUN?t&@dfqP?YaO*@g2CojGDaxGI_Y?e%
z#e%{3#+ARhSVP^*g>z_#oOJe=k9g#kz>C5Mv;A5HlJB8Sh~A#gHvA4mdD}p0z42YL
zZ@d>O)K0P)$HdI7yn#CQ{YJ_B-c)HUn9jfc(fgWz{8zsmQmGsQ#+EYG^cd0?*~-Ij
z()hjg{poud$Kc;-sPDUyMpz4%mg+caUzdYZmg7a=>IkGMuc+rTdldgR;Fh{oFr8P-
z-yW!DTUmf$md4=OtNqM=j2(sPmC*CAmx(0{Zv67@)X=3#0x@sh@Xrfb>RuRR83P|S
zL2@i)DG%zjTN16y(f7tBBHIw|u|kL1+f{|1<u|)Cus~9ILu7Wccf!`*8@=^6iL-tq
zDV)io)3Z$IqlxIv+Yg|o>F1cW-7Hy?br_Q<?_}9?$C8D!A$t63U^?}eVd4IqE3G`k
zoTRDnw@9axmjf}}Lgc(!eereRY!uBHj=4`xlf!-|nL*L*>`&ih!V{s+HqAOob){Wk
zy1|?MU6=+FbsutWJBmfJ9%Oy89W+NpTA6%|hG})>m5OKZd+04#e3O&9dGr)M$iXb}
z%ObqgD5bsu#iq=lv$Sz<3_rEj2iYRecjf$KStjPO=$_}u#lZvJH21Ly&z`fyLxXAF
zAbAwnIOBm?32yK5m29{x?qb9JQF2BrD>NY(S2o0Q|KZ1QvZ{$(?bpC<QvhOa=&|zY
zH|SfMF;r&>#z4=1tjoS)`ta%xTXuLoZi;86;Ug)#z1o=!%vYkgJw~>PN+EtHwEN|G
z<czDJ*bhz;{Slhd!Uw?=K57-?OP7<b&0ijHpqZbYxQEtO{gWib?Ps*J5V>i`@bZ=o
zsb#4#+k-D8a&`AH?6wc<awG&wF`~aeA}U+u*G_in<#XzL<O&<IDizBU&tb_%wVdeA
z&XnQpfxrD6gnP+{iWJYX%&saqe+&GmVp}66Ab_TMm-4C_AIXYG!r!NtK;ulp5xO@3
z-&;hUqwEseVjf_n?+&pq6`u0_3uw-R0QS7L9D(t}FmID8lr%EgQM+<xklckXjd;hZ
zbQICzpGb2X6JW&bfS(P_L*o)FyIe+V2Ax2+F|pL&QB~Ue{u<tGOsS;(V=A7dxllLX
zH}v7HKPLS#C)<Q(s4Wtl+ka8$bYLb0Z@9(vZY^Z9PuwEAfeY{yZ;&<rG~#y&4&+sL
zD)Pxj+~7YrGT4p0{;fe!e+j16rc1tWC=$%A0{Yss5Z*z<Y1bh!*Elj7?K*9|V)G~-
z<m$wnZ(Jq6%6xRrY~zMwXR~WFKalL78%=s6hf~HWq(0tQJUa&BSl)N^%-q4I+a#ed
z#)X}V5e%s%J#n_vW@?RhrDszHqj!sDj#6VTxPKON)%w6w{U6DO-yBT~JV%oH?p3_w
z`Y1Llu7Gzxe;@x^4&nHqJbojfJF8L>eurgeU}v=uT}1ME&3sKsWlk+WZYS>5m(G!|
zswtM224O-)8dJ>^j+?wwFk3If=9_<I{qLRPfws=r5Oj*ZjNM6&f;L$FcOQ&5iCM_`
zL$vFaV2r8|QymeG(cW=TituB#vM-pWJcoaL<$>1=TqtkK1*XTdC_cAXHn_e&9k4n>
zPcQwE{63UJhd-+E!ech5xZA|m&$=n*bO%w@cM7#7{Gr9lv&muZHwsg}ha>sHeB=->
zs5A(#_|jTrJGdhD@hVbXa)x+KE&aDZJOi~G`0I1}WKrLRinlbPzs5Z}zeYyK{MG2^
z(imRTUo$6Ulwe23p5PH;XWfvWCW%Pj0M|wX(_RiLId|lGNp*g$&z6jE$I5rvv>@=i
zc;~2M#m?T)Z`MY)L%BE>d<+Zkt5blx=py|nq`FTN>Dr$|boRnS?(FBnw4--W!n!kP
ziVwqqKD#iq#|kV>x1xcQ=W^+jQ^+&jM1O9MV*ABjMd9yCBt#igdxV!@C!gc4xrg~L
z`}dfC@~-fbbSCek5wO^I5>vwapx4q9<Q%pZ$3HB=)*<pzg>z}L%9#;#%KHViCN1OZ
zEDC7%L^G-!zXlnJ-&oPK_iUF^B;qG|<H_^~^dmEz+@D4>NB22&bX6~yHSb})c|3Rz
z^_+y8rzz*CGy5@r6pdVN$v)?)liyP(ZWZ*AhdL-^ADntK`^}PcRI?0HZrZ|zjU2~@
z2By(~wez{whE_Un`H*3bE@rEBhHs<?a~p4tB&+T4>tRJ})|RlFvmUW7ZmQB@j+eQO
zeJ8Xnt7I2zZcwy;VRj+wE14gn%d>urWp|bOQ|YA$8W3_)^hwXNn#fHO2az?jZcxVi
zf(|CPO&)m*KI2rTV7Z3%Agd53%sN-V?_XQR4%)vIta4+l<iq&RNk5s#=CiyweKNV6
z4n+NUAMD+@jGku-r`_vs)XAj>?RNB}xvNDdXU&7DMGwpn<!OW+sgsfXdJ6v-M7Wu3
zEt}R|urXAtv0=|)^a@#kLGvC<A|L-|k~g|ka6cbe*N3n-G5Mrvlh0ONJ-|s%0VXFk
zB{L3eLH<-PV(tMrQ?Z5(4>F_G-II8()EJxBx)98N^P$pbtY*F^_O29+YL7CuX}ps3
zC~l+o&;a;#y@UOdrKmQIWLJi_vg%huDR!uQ4zG?tXSpW+Q1n$6k4T5#=Up`V$9mYd
z#=@z@fY1J_A}#AK{=O+H(%tX6BIUJQj`oa&sMR~e-?t{h^3!JQKiZeu%8g|9VjtNn
z@F(kWDh)0BHL$VShqOem*ZY2(X|(z;E_cI1WISW=pqC$NcKpE47jo>QQ5uF+oWYAk
zS9X1^5$2CENBS#2(hVBO9;pSO(|@Na^Qk{Qc>h##>0BbaW$4NjRO=`w=@iT#6tR<0
zSxjNk6N$o!VRUZ79ht|@eysT3d&w{N1vu;y!>?<4lG}kh&=mW!y$v5({O3OSntp^V
z)b#1}mX%zo%Sk+LeG8QcIqAzq8Bh&U#1p@rsBJoo=c8P4$F-VYR$WQaEg!JSRj^`w
zuOXh|(NwmOc3&IASG6CbA^N*vGD?>o%VwdMwh81LL->`e&$F*;t75801plG<f{p55
zh}}uWSiEy0sdWrx&7+*?&%x!~tIH7jaV&}DB`0IYZZ|kQ`Y6%%ZiBDWDBh{;5m@O@
zURY6p^12<k8gNBs(mIDm9Bg9)!!zM5dxP_boiL-qgXf0E!Tb9+cyDlp?zz5b-A2&e
zlEKG!Eo9nuJ7wE-?ukxp3Ez0N7tOA!5<cc=%+1k3tdbE1cW<NzI!Q2S8V$MKm+@xd
zRfJ7erlIOHkT!G%?B=dxstV6!Bc!pY&8}hh`j^t?wkVn%G%(x1Hcir+e~eZCH<1mn
zdq!@JNl>{GfUwf@Sd=K?7bj&Sv}~>{eD^JCwHgQ-7miKu7qO_IWbqyufWJ2uQe%@M
z=1&j8k43TMy(uMo`?Xtmx;&mlI|GTnXSiO1t87le5Nb{M&h3Q1M00&Vk#8|XT)=+z
z#A71uRWD=nPhLRB&ra|T`wgRgB1au3<r8N=la&3?hKF>L<V@=}sOrf{tNxy%5Tz0L
ze(f-OJBTQ?Qdt_>ZjZdd?xfIT7+vh2#&(;JC7)d*5&giE+nWq$O(9*$AxE8jhp(be
z!Fw>gF^PE=CerFo>)3y-uTVR=3s*Vi!8L9KA>*@{W7jMsW3@>7aKZ-tbWP~U0z=s$
z4LN@7wI#D0m_p8aT~XF^KE0ydbok3+%pAIars^Fe#m9XRT<8Ib*by6uJ=7V^@$}*I
zS!NQ`#@b$_<I$*ANluj$bO-v9WTqB>)$1~4=pNz;dT)?@wJ%M}5Iis?XXL-rqj;NK
z{;QnBf7V)B)5Dg>FRa1bx>Zyc<%(ApinzK{k8WBgqD<2VN2f33F(Q+3H0B^;njF~Y
zBMV{od<|r)RO!B@@O!=ON888mp>}&^S=r3rRH!%^e$|5c^(PAx&d5o1hunt0)P_1I
zx8RubOLUr>%I=4+WE#(}@v6iM7CKp;?7Xk@74FvT?=b@uEiXrllJLIO?qeP&(&?X4
z0gkU2h0QAB9e6VkTW^%ppu48@;Bgfy>!$Ovy_IMYecSg}E1Biwebm&s12w5)@iyI-
z?%F@*8I`NBBHxPw702^~Hak(?>#Ah_6TvP!lt>fDeZ$@@cC5N{DIWG3$LzN^)1;bP
z2ww9FPsDk(?f6H@`-@?M!4t#gMTS$@^^2r6-GjKf4&1GWut61q8{mDM$Nd_NjRu5S
z*8ZgHR>J;UHJ>CG7DDE_gnzWMK-aZ5>DT&bc=Z1PyR3@{OmXKwSN9>S@0Ynp*RAAu
zKLzzgB{-#U53~AxApL1;>C%`q6n>wKN^gHQ=*m-WYyXHv-Z4bg1qq&Zxkj~pH{;4N
zF^5>u0|PcZ$HF&R<fs@#JwMvg?DUyzh9n0sSNqZ)pSQHPe-fsh)25xFA0U6bhxDA`
zK<cC|MZnAB$WWa_u3kO)S*06DjJl7z`*i4!^(q=>par>NLu}vsT$Xg>oJ><Ko1Sga
zhs{KpVB8NPr?d-rINF?V?eQA#mc?@K5Phy@z8%KxH^kp4ya(?dLQY8mTle0@=sa~q
ze~F`L@kSptVK4o+#0woq@_Cf=V|+P(9y3dZpuNj+YF!)zJMsUYV{U||$z#|X^M5q2
zXgQZ%@T8_`F-#hGQ*aR6vD_;cz2mlVS=B@w?x^C;*28hO?FOB{5X_o>o|Y)uq|=^z
zU->xABAVKIo`0IKko|nVirHK!qs&6#_!N6W`A2{F;}xD<9pBN}B1v+${u2BTZl)f~
zuTk4>DGJ9tFwIcihi~~qSm#j_x#LwEa{6=zUfMIU)76-QjfeB{Bo_?ol}XxLFY@h~
z3+Tb02yE^ei@Rqlsj*@bw##;-%XB5S<LYX<yXQF_X#UL4O8c;sUf!5ey9ZOcT|}FY
z=p7fo;Jx!NvT-3z>|kV;q%r*u>?X<O825MM-up+B(mWR!DZj>}-O1>0p+!eqe!}Qv
zKBb=;E_vI}3thhrrZutYOt+5`o8T<EuL}QB!`I=o^Z8!9^HZbAFCQ_(;xs(&o`ROE
z7TCJ^IQZ!ERNxee`#pv8AyyyPEmqOYs&&+$J&l(AdJO5C6q&WGSg@wsu}30S)8;BU
z>cc#;haS3(R*~=NsQ=DfPkZ6=a&5LLG#`hzS|U{ag6#7kJ(S#>M^hx_nBBDmDLX@X
z`+-QTJ@l9zH+;iice#x*9sz9Mbb0#Q-41p}k8tJi9-MlrNnLKpOJ8<Xq^8ZAC~~eT
z6LSpKVW%oR8GM#B8Vk`pa0%9m{psv!`*3c04$22rWOuA+VW)pP(~$~!_{8XuwqZJN
z%G1HfkTUYqHV|j&4Ei>+ij@9NK(|>XZ1V8Vlz$-%O}L5n;r-y;V>sgLvm}3Z`J*Z@
zolM`2#!0b1&pdZcl6q)9T@0<4Tt8BWAw@;#RpF1WsgZaR{fxIw?M=n)rQ9-UC@#k*
zQnQ}1$ie;29^sPAhgjG0m5WZpQ6U++>t}J*nWZ>0#1%R#$}ra_0BRqKdE^dNY0CuH
z?9y*h6lfWa!}&kR=0PJ86ZX;a`9n!zl>|o)eL=%bH%av`&MU?KNnP;yyJr;PSLRi!
zSZ7Ax5(d#$$!+R7Ur+3>KO^Hfrz`F)lJma~F_X@Ef(fn+joxvTvpSdcZvKEz2k-C^
zMY?3O`yiWY6~d>j>yDP*$+*%>3q~(`ko(aLoH;&%*X)=pnV-4<+uhS>ZSOn^_f3OA
z{c)r?<?=eS68d{q6?2X<{M$E}HXAr%z*ARPnr=iSiG1;tU${T`F|tHI;L@GL^kt46
z9oKtmTK(XLaKCFw7iq?eeNnFLe(`LUoIIa*aj3!VRKY$jSW2cj7BohhMOD69SkbN|
z?QZ&w$M;UdaO3rK*?BP3E`F3XH1uJ-P#dE~_h)e`r^OeqV95YiQoE2U@>o%*5B$fi
z<0Z1~8;y~@?Fjq6K!<uxbTwUj?lL9czQz@A%*Grg@%KCCnI0SyL_1odkZ|-S`+K@C
zrF>bBgLRIWG$xH5`FV!Td%S_)&G5#Gkr{ks#R;0KkVbo*ttqPQ2mbbQ<|}q<!+TgE
zrmR=b8CEBF@2gC>=T<+Qt{zVYp0+U4KP22H*U%>M#keOSY~k8;_9gifi!{2%3|H-?
zInRX0N%WD@Hin^3NE*5R{KwC~3`h9!<FMHukMe&vA>FEyGqp9Id8`{KX3IbLlUWLs
z@u3?^%GPonR>ih>dm+z1n9hdSbES|NoVzrEh98}Qx?zXW$NIF`1O4Pb{;Ffmq~X-_
z)B|kVd7rzN%afmAqW3$yg47(cXvR|?gwGm^@kZw)4PNFjI5|qzGx8@ZmX2m)jKa{m
zB%ft{$`f6t2+T_LBP&b6BY#08G=HphW*|k)xq#cr7x>$`8EipHFr3_)n8UUl8o#(z
zlJegnOp6KOnHSyheUUj{+%v<tF579k;EQ$9Ze-Qd-5AAmWRG~^#Z3=eqVnixS!Sj_
z&3%+XS8^taF7b8NbAn7#89f@?GJfNIoeC|txFU;i+K!_5^K8TTg^1m@8k^RBVXfkK
zC;#0SZ=7V9?cj#$3(2rn9fQwK7g$}u!|a#eL=QLC3$oEN;g27N@3I#(Wq%);Rg^Xc
z4}C33?0<+Ke$rc}Fe08}Uc}&HXBWI3^^DeSIf?K!GFDXVLUA1mY;r@A@SYd)y7&5!
z{Mdx9<FX|ik6)pY$NF=>p7E6Ocorf*{Rd;SYV=E9iMF$6C_d>0_V|^;;pcD3n90HH
z*-<?*?DrANA_mhFpHil4{~scBu3_G>?i9N<3<hz&^yzf4L}_m<?-_gsx`F0mpD~gy
zM*pNe5$&)$xrg8F?MwN~24JYH77Kqk(p{-C*L)~rNp>Qms~E{n_5DryBi5rlb_kS=
zOX%T&;Uu#Z&Kestd`nm7Rdp7mt|Tv&^;bq8jWB$5+Q9WMd0}|SG}J7eC)rkNfg{)Z
z(p}r<lJ}Ra5Z%*X{61{69p>L+KW06H`{*;MO%-{d%3JJX(FW?iqc?v(YZCX4DPsXO
zn;@w+pv;2vu=FhzS&j@cEBMT>i+g~*oj;xMaG~}d9<c0ll9E>5WmBJArX0xzKD{Ld
zra|{;p6U+qJt@Ni?NqFgpTH{jY4P@^EZEzCPn+n*T%23jno0|H$NUMoK5xh2@Ic60
z4ncgI0}J(vXM^{SrdK|_ki5o?+0!T4yXBoB@8k%FdPiQnXfM5z8;V`G=V9(=ck-Sy
zkC|CMMBy$)X-1(h4BMjVk)0L{v)n2Fk~b|C-J0*lC46D*LFg74)89#paAk@DVsC$v
z9DT8e_Dmkct;^=(erOiP_8Ey;>MA5>p+dJUR1mL_3&-2B2(~&Rvy*k9sU~k=B28t-
zoQ-I~>Al#oRfp2o+@Sq;!}yxKIQn%#n-ci|JS~4sHWMGg>dg~$*D;o8Z8E`v#C4S7
zoyW@0hT+naN$eM{!A)l>s@)!N^NUaTVGZJTj!M#nFK+RXuS(dS1$SVV+!c3Ut4p;e
zwX%WMY2*`9k1?LDSd+%ti+#(<T%ifZCp75o;6aG+T8^YKE3s2qS(-m=Hjag;NT;SQ
z#%Gx$TjKbW_v^kDfulLfQ?~PAG7b9MBcA5y43YR2e_?^X-B_>1&a~Xj1qQ<>%l2li
zl>BY!!>61TZZf~U_#^zi*KIpwmc31|q31gEAMy@^Le7#y0>{|NA-M7X&sMgCdKX6{
z;9w&!KQw@>hgI-nYa}$XQyI-Dwu6h&XYS^*3f^_6(VF6c1dX|5?HNjqUjWs-<&@Hc
z@cG+1wq(l#{MxS0Gh?QcQ*$zP(iB~K`Oea=&4c-ndoqMyERfln`I1+^D2mDogr-sw
z?wn{~Q%k<%pvVUWIS$0H%I9P;Y6XS2iE~(nDpGte()^W2Bq_oTWZ?T6d0Xco(O?#r
zTJPY)<!WUA7AQy`MT#ADb0aOP|4IhGy7TK@fJK8}WY2B(;5`d|v)c;OklFJkkB(Gg
z*WQi8lwYZ^dU6fHa*;fG(Ga%lYN=rPSHMJPEaB&1`29Tv(>{r`<hHw{ZvHuZvv7i)
z$oW~F3#XK^+vrE7;0bG$!dYacrwr_bA>PN3+n_{CzWK2MI@8d<DvCb;a6{Hg;fs2B
z1JP4z*xWgnNqgaLsIH2KY0CmucP}3kmPJc~ee&3Y9#1f9ye+HldYuL91(N@zV(jd)
zijFS6I(16MUHpCaku=V1h4%OfREc;f+zjR8bZVKg-*9^U(G8EMp2EVV3(>J=JMXTM
zPLVGiB}dGHNW(vz5(aNWaoBTAnIFJ!uNN7{536}eu7WIS*kgqMI>%4)+qik@5BjG5
zlcYF>$pY(>*v{SABpY~`jQ2;et?l`AW!_&FNfVf5wFdW7cV&`OR}lPh1+*I)c+y(+
zoK5>Y5NWUuhYQ<bmQ;c_9$m#=@($^|TaC<!-rQB`GxWr>2Aw}MtqFE~#XWauhaH5`
z+bnU{x`>9h)9iCa7}R!Hu%C84G2o6V_Bb{O|9lgTc`yJiDrq?NCJ!IXKBDJ}w-PRC
zhf1vqmfs5_&+@0RUO5x<%lu_OXY_*M^nMiGK9sGMq+-Kt1v;^hV@v)dI)8j5d(`49
zx#*N3v$&SQ3S*v;`<!U9o$QPjqt7BkX2vIoefgLgcY3eg!ZNN*6g*KSyb!!%{c9UU
z*SBlVn=9ifux}*3z6?NyVl3-iWyA*eFcNo!Hrz^froF|UxUVdwlp#N49~;%Ar+)m#
ztsWUP*0u}fj(kX3MOFNs#XyvlmZH7AglE^Kqki&q{#@LP`ltEPs8O-hE$J%FUG<$6
z7e8m`!tcp8xv0ZnLk)kKt|6^j7D5y5-D0Yb&1h$OHB(97#W$^M;N}oqwb(M;J2M^m
zBmLOS1FgK~{bgR4SAe72p3wQLm(bczoI?)CNz-IKaYo@B?ug%?|CX*WJFST1%aQDv
zv=Y5@v#_G%E_LWCAUb9?oM&h9!?(QPx;K#RzW+!zFXIm8mM3DG!_e$I%0cX)=NOC*
zmCrFb9>jbr2g~|9Sg>G`gFX_KNGf?hdDg_I?Dw4C+^Ls4A5^Nt+dcjv|EN68D6&NC
z7dx5B<S^4%opkcqG>8O^4kPT-S!9<c!M`qnx7=rJ4+x=8dS6)mk1O=wr&KC`5RZbJ
z!}zfgXK>yq4vxDvAxJNhkBbo5hB*d^z8C>(?`}vn`^=+z_ch&;oy+Gm4a1_VvH0lv
z7g-x~K+1|tq7zH)m))}`yoy7c#RSsWk%=g~SKMx6S2q7Y;TOz)z)wmY=<Hf8diCoJ
zmwT{=tdm3WJj4e_9=D^lQv&b&(*`eQ<&gI+|Lm<RpTU0dR$O&_f?P{^_N7yRWY~gW
zW}xhWmUYjVqv-qx#$U$#)XU^Ix|)ji>&bRVzX@OCHO%e^N7_|uO!nWw{`iY|m+L>i
zbnsVHHm?$H$2x>|iAF#1dvjecyb4OI$>gL4eF^PG2WzY(Dbd2^m7s?)ky*H`)eX7L
z{%{$cL@ulQ!eYFt^jk;}ber61`e03{SR|AB_n$mrqdTc@KFNnE-(+UT5^#S>2Ua)h
zP}q(L(OH;8zE1NnTI$73?!MycY9gn6t_<e&*J;{}D%?5Z!hiHHLG&tD(|gsN&CQu9
zyJNmd);8OXem7jln*+A^+Q*D7ZhJs0M9w-pqd%$K_Ldw~n+&hOJJiWNn$56Vf{}@_
z^gyMBGA5M4s-qm<^S%qdyAqTg1E9Xsoo3p)3J=s?<`HCq;oUv)QBRNAES&)N-^ob1
zF$4PQLAc!hi5V`dht|w1Ky?Uj-))N7Eqf3Zybjj$uHj$u4W{>5nbd;CJ>z8t_fdI5
zg@-Q-hSfotLRk-}Y1i>n4bKpH<0$49zT}a4du6lNZ$qE%8tAwt7{vQlvgVh0l9o3g
z`G*u=M4x|Qx_(pzjkONog|=0Q6W`%kV#XU2c8cD~uY+#xNP08l2pVSRV0d#3@37uX
z``h0mqwTXSSl<DsMD{UK_7sb9#$~4+c!}|57wEA1WbA!=kurS0$x?mAGwkL=*^&Nv
z%;iBQ{*3OFGuGXN0_7gUa<O=aba`o7xH^_<OG2RYLWTZ5a==23OR!q^8)*fXu(MBq
z<V753bM8)}QxT(3>HMDcIKL1{PuJ3dLE(~0wS2MD8i#$U-<bUe5AM3aiN>VQf#16f
zR_-$yBTWcPzdhm4=U*TP?d_&_7yQM{Z-rPqC73S#{={9b2eROmous#p>cV)v*fBfj
zWap2&2lMNf5aO*QU0m7~zN#Z7bHsf+Xu2Mfw?1O`jc<|7S;ppkzmAV*y-?kCENooM
zn2xU>nrq`Q{i8oUf44&DwS6&lWi_V#>xIHR!712~hAk5cFyTTLS31#%F8!iu*ri`|
zvfE7RzTB058#=@CbScb?MBc;KST^_WIm$Yfji*byp>AV3En0qnK0Np0n|sY8-9Eo1
zmGxP&A)PEFmG6sb=$<2RRQbSa?GK?_#y4Pd9)2FF;QhY1AbQlbsRtHLK)g;kZN8Y!
zR4AD}jk^e+Mq|-Qv!nR)YLc<S-S%X)0S-8vm;EZA4pYl+m}B_C^iNt4{Ffvl+%X9E
z5KK$|xU&6ibMUwaM}A!i**jNY-QNY+3E`joatmsq(@8gB3*z>2lqt)R>gmZSNO;GG
zU%0>*_Y%85{TSXl`6SlOUVws#Dv}Mj$m|tmxM(wq?#d3c@grgo_(k~STyB$j;|cy%
zsevx$4#r#rO5V=TK!D>>tcf(?k@9Bv8r2VT_P?aa+bYQUAo3&uCAcEDNMdxP5ZPW~
zwDozZEILO%N9orEWMBS-*n7p8_C+|3wJuVol?-pW76zJrrDw-|S<3BDTxlzn=^UFz
z^QOt=RNBf*ndxiTXxQ`Ni_W7>BNZ#l9O$%om;75ag~Glxh}qgf++CA};$EKc3h6~B
z)~C|UTZ^f9+$!3+x*0A#UO>@)9wY&7_}5Uuq?wP{qS$W|X@xo+Su+tU&K;vUu6uD}
z_Xp<l<Q*$A7CytHp)}rR8HOn8VtH&VH2o~7_m>oOw;IbFi@oT^WCdyc(ia$AE!-!{
z?^tb1JeQl}!Aw1mW3gKu&687;)`grxR{u||<U>pL5o>#BeA^)I^bTy{FwT~i+sZCH
zE5wqjSrqNLldhjRhOb|iBU3OF*mrkWnoOqQldDl-bCpPb3YHgbWUGe_;o6s)P3zY0
zz>Yj`q<-otz1A*VjX&d|(HKGdTR%fHv^%0LYN;YHft(||NHafFaEI0mG99jvGuo`b
zO!Gb?{dGxL_(_?ReSEO=M^{=!A7m3cD@*&DPm|5qJQsB|m-h@)!V}ddw!$(R&GYKH
zO^?%jO3Yl^lN`h%H3Qhtg|7VZ)-(j~JWczHFR+VWYT?@L0P`NYmes#lMQgmrP}1PB
zeEjQ7^z}BCOw$Vzd1e)ii-_c=3;j{2m`uZ`e?a{@do0hm&MOK{P`1HIoYBre>sAZ2
zThm!<?^$$1!BM;uZ$NE-DKk^nke<!XVO^^a$m~0h6??Ot)VKFm*6XdvV)Zq`7AtdD
zrS-x!V{iJAm4%Vxg30~IZ7yfL$uuuxAO_AUMcr#P=_<Dtu3S-yeMXkh=#dD|qr-5`
zIh>oc#q#NWTKQ-@k%P>E=!tisC40Pae4szKAL5O9W0E9q_*049{BTmS@rC~_H{MkF
z7?Wm|@!axlWIxr0+4L)rJ(djRjnmgr)gMo&R#b5V%^;dQ;GW=Ctj5vxwJ6uUM>Q+w
z(pkHw7`Wb?swZ}*NxsjSUV!MDmL6tneFMmP^JALvawtU{%OkTjFDSv3qw!NZpH%af
zzg94yUAH^Yuy=ZFhDrdmwnx&{rD|B9x|?tP=m6Jy^{jsSUf%j!<Sn8$&~Lw7j2*Te
zmAg{;6ydXpyCCjL0gq&vljhN?J;K4RxFOqY(|2C16bxy2F?-qL3{~YSNez=cY3U$u
z7*4B#_UOZ0&g3g!p?8|Bx^|UuZ82-UD>~VC&+{w0f<&e<4u4m7Mix2qQAWWiYI})E
zj}qABZuzLuY^H*wpVX&r3QlfT$8<AonBD)uUdE246^k->+<<u0j5WfB&0py)zXH#M
zJ(TZTiG|M96!uj{t8CWM-ZfHuICoO;eoA1zG6vf6zbWhKHXIB#B8MHX=wa<Q+}|yM
ziAoB1iB`_j>I&?z72U1%kFoGu1tyE0yPHNDz249Xxt^bBvV${gtFK~fRCZF&*V819
zGdw6}fFId9hM>jbg2-X2;)hEN2KC>FnK5ywxn0k=w2FPvX+hH&UoySFpTD%(BKYTz
zkzoCR?B?aL+reR^x?RS5T)c@D>mRY?7a3@EU5=C~CfHtm2U;K3k)O^9Y7?_J$Ckqq
zHF}5nSzj@zt^w!&^rPNJZoF=Z3aK^A*c_KfFyAnPw$`46<J;%_^URmb%}E8pX8zPS
zLJP@B<!rmn7tHJCCEN8eosF6#!>F?fF!RYE?P>>EZNYWSq(Sgmra=RG^v2FaLwuVv
zjBd(wXvN$#%qnh^jT)5*<<>j+Gj$bOj<+&J_Y17{BWLQG;@`VeLd`ooVdVH9t&|Uy
zS@zvYU*i%@uPgpRd9GSch{r-QG0kE+a>~p*>=UyK@6IOOh=U+8i2m3Jv{c=}_rOqk
zv*izT?rZQQtwU_k_e#`?J=|MU@ht3rhZ+=4BIt4}-Z`#8?To47+*gY)4KML!W`B(M
zXGP)De)53r7NWOuohe3dLt48QOR@ZCI$S>tPxjs5TiPS>yi^IT^7&}AzbhLrTzGR5
zdQh$HeKyDLmn7x$U(@52u^3<F&RRA!vP~*cvU9~(nWbg|^sSY1!ZQAGzvs$1p*rtq
zwA)ks-FplX2{)-x@NGiQb;r&Pm*6{i1AT3N&tvabG7?<0fr3LF`05CJ<wbA#Pnc}n
zj8=(HdmfuOp@`StFT%jbO8COFn2GCTl6kC`xK2}(WY8<T?c{|uefB_Vlf(yb&7As8
z!l7Y%1WycADf;*dx^l^%jgcQfRh_fh)s6pnb^cE_Wd3*V9<IoZN><bJ>gOohHiXW0
z^x>W}3uL9TDj4tbrqf2Yc(U;-4tLwnAGySl{+xQqb&i3%Y8t*xyMkqRcF|s^2%<Y9
z5!?G23y)cd*Y=m`Yo#jO%<HMy=qd^)%1L+ixK45_ZnHgO`(wh=X|nH5=8}wW*ICr4
zd=!5xLB_xxlHV_a(0z!ntWdF(u5`~wc9R114Ykm5W-pCBrihy!`KH^oYp_r*i$3d(
z!LW|Y)IDMtran@JnfG0qvFQmE?d*{JsR!F-u0%?MYw>jV8(GJT6#O@G6pRnM(y~vQ
zIazMrbn;aQ6>nGNCwk{&?}araS2b`7T^_}*Pnd`fk<pl1evA%;ZlRE$Qe^vXM78p5
zdaq=P<`seoHCTMt%HQzo*&o=|#f|Lhgj2*0Zl%u$4w9*SG7odg6g($8nlm?%dd6R(
z;^LvO_ZdO82AnN2%%OItIC?Sq9SZ+*!Wx}i((+BjBISCzzUL#V%nNax2TEF3-bBuT
z1vsx_gB#6B6qxk|<JIh`uCEokFG)a(Iic}|AHENFMPD&XrlM@JDlMQNS4yd@;v}?6
zUSQv#nY6O2s&wRy2)^^}6E-%+4N0Pdc>jGcJ>PMa5+9jhgq|8apALq}S3{i6FK2@+
zUP<Px?1|Os8Ti-#6Ee=l;#avHu}4QmzbTTon^?kjeLC;m(-OfJZ=g0s548&yz-sy*
z49|(cl(P<?zs=~V8cRLQBP31o=S+J){3Nq*Oreqf*RZQRklDTw_fR`e*<Bf7t6w>@
z?ct9F_5oP*@-!Wp6+wq?&S2JlLlOUbhip@AJnQYhk<^M$!^TpLKNWrXcfUX3uVE)T
zWcE@L{lJdBvQ5B()Cz8_VMTsH!Y7^-Ku2EZ!n86Jxyn8`n<Xdxz|3U7{ywJ#krUBu
zkU&#Mxnb3IMpnIJBpU|JLvT_7|8s8@vLesJ%Ks5$dsOK2r7Y6gWtVL!afG%+Dd%KP
z1OM$L@+%dE^l17$3SYaJdA&8kzV`uuLn3uP8-feN#o1}kCR%s!8Fc%&GDp9IC?6Yz
z^aCS#Kiz0vU@=hkx~iTPcil-fM@LDV%lA$3dX&sE3*X6>8iZtj-C@JkWV2bJxxcB-
zG6TAlr-iMl7g<BmZ;@TT!*+xy@~;1xz+-?udB+!E%ft|xzkDd~s1&@w&PPaj*>+mc
z<0ZnsO=a^GizrTX|LS}9Ldp?O6dnC4`5RO~M$OIWp^*l+{!c|Wq>MuMR-w~7(eqfl
zmkJK1&;<I;Ji2)j{}e|@WkX5FFOO@F-7I{V?%Z~KL6(w>FD1N9L_zaH@~g?Grk<U0
z*7f&-hSELc+m=yAmxqw7%Yb6^8}4~%8Q;4z6oWS9(q+jFKG&_9`FW&ab+#&H_GXf~
zVNU4O6c3-Px3ML62G)!hJ}2HAb1r&h-@H&NQ~z%nC4_!O`6F!#JM@?zT@t|b20UWc
z9&tRbe>?ZE)1{2wkJ!@SWvr%Mg-qU@MdN{fuwB%db-v(5iti<YgOrB!j@29|GBM7~
zm30~sAbIK>kG#$SWRW7?@sE4bhz5IPbqs-F@+xjf_n^3SG^O1ff)4*MbXjbK%$#9@
z&F4lL^3!A!bdo6luN=FgA)j-#YYG}ib|KH#b%+pMfX&lvAV13s8O`!E<zEt0ADc~6
z@@JA;fHu=HJcihglUTU3nty(}hwj|g0j-RY1^n1XZGwL>p}RhJ%z47>UN;L@y&{_C
z^b|#xWstb3(GdF&q+3_PG~ceks9`PmEI9^M6*>CVkjoFIKjbaN%O#1^ow3k$7;5Em
zc<#|iDlrO|E%tH4z+)Qhx99>MO}dJgBqPm{ACPrT4hp3|A%8lLZ~o8;tu0!-&gqb8
zL1F>RcD#-3*<ErBE^X&?58R@(#0jjke-ZEdNtFkDy#kf}{dl^0ENOd96=E(W(VaYw
zMK*uvgHJZkE)~3!jooqn$Ys8MhnvKG`F1d|!%_)K<fC<b_^5vxWM5&xU7Ht6Dg*z}
zl5_r;T73qachpG+uTEpRcOSBWCl#cw`ro;9l!S*Sv?J=~CAiG$O8t81Q_%iDNWNr-
z7Zc~PHQUPZ`N~MLi91Z&^E$9ZwX;-nRx^)CPA2oO!LswI>Dcdio1C@IP3bxRGL>Im
z$DhBwK!cuN#_FU87(ROd>l0$h4qRJHFLPc)yK)Ggp1B8y#~uRq@{H=2q~frm2cq<j
z(ZYlu68q*TEEpSy)CUm=Z>Xh~%-xjt+kiTH%S%l>-=XZUAP#g%M77FR`fB6Lo=n@$
z4);93Q_Vj?y~Ljl3n{|Vb@nnjgDkQyDqto$R#g9RI-(y>&2H&U<R=@#qayq<ZKud4
z<xj`{wEugl<Y{BzeT?tW!N%@o5_tc>w7(qFW}lZ0?SBIH@lW9%B$(18J)o(4iVXhE
zU^hbLxRr%cPSL9{lsxl5R-asqRJee#x872BQ%g)Oih^$TH7pMvi50KjGP5ivs#~5)
z$!A6*(P=ggr_QB8GxEsdYZ?yCGeVp>qrBV{7(Z0tQ#Bj;faHm|_qG5V9jsuKtj4>F
z-|y66XL0^-BBLEH6i}1GN;Wx3;`$~bK=i|xK5@bFEHMWQQDp`r_H)~Nx%4IW4?lSL
zDiV}zDctZ5cBm|+_rp?Xte6>SE6%45!z(D#PUbps=Meu@N{REMDd2w;oo77O-y6q8
zNs>xS($W$tsU+_6xwWLJgi3qwU6Ld_Ns<Pk5<*r;e&RltQAP+!(hwn}p;TI`|M@@Y
zkq6cNa-Z+_T-WFGe*acRhuj1Htamu>Pw7t8lRnFOE5BqCHDBu3mW!EP=fTI`pO2j*
zII^#=lH_kX&OHpr(CI%=WxgM^R%>u`@J4bpKhAwN=;670B>esymc&P8V%q$Bq&4I#
zulVjN_-W-RoS9FGs&>q!DFuhNj^GBnuA|q_F;p@7s^s?y54Q00S-Pjw70+CSk8g4S
z|1)V9ma7afdDq8}G=yVo>&qf2ZV3Z_KNlZeGnm1KXxW*7JCdxWl^H`1Z-;wuPkJ0G
z<}w~P$!1?6_H7zL<7U)ghT|youy;ov<2sBvl*(`H5oc?%5)F7`hR5Gr1@kqDpSiV&
zzI|+ysr+k_ndz94lHY1cqlVziGB+mKWGD%}JYBM{Uy$gy&8BoQ<Jyrjh}_eSs5LK@
z>{f(Jl+~lrAUxooSQ_>I_840yE~Pa|y)uKpwV^$H5&wBG1Op9xC{*h#e~X#;Hf<C0
zdU(mCGPnr?jt?cd0sW+h4~|8^YbkY`E$$YMoscjhMDlL7KaE@aADNl06my73N;;K_
zhWEwPO>R2Lee5SubIOLx%39V%^#<P{b~}fR5@jQL=CN%{axzX#?Fm`<AFkZXi#sI`
zM(Cdk{JrB#9{-Y9uP#H$_Czk8WvnLS;Rm?ww0AIHC?~bbIw=cN)WoK=b8z@*OMQ$V
z!p}LByIQpJy?Uv*aN{ohifv<s2F;ZB(;W|l7wL86ee5f2=QCUaDCEId@=yHDJ}$C{
zM~_e%)}WA)F`^Krf81$&CkZ-Ur1A0!1>Wng7FXH)PquZ(P0=@7hva}=)OFA#L_7^5
zi*Y`rQsK_L^;M~GeLYS;KY*k5_h9$F50h>!!o0c$TG0H6tKarUqVpN-?KY0;)eeeG
zSQ2Ej?sHwQ2~;lFdtoaKxsK{7@;bGZ`BD=0Enk*Vzu`9TwqHscOlx^__XSKvJt89q
z<0(ol34W!<gu<_6@jxy!!q^2yMk&Y=IUAKa_ly^kN4)LDd#WmwAZ+|rl!}g>!rjfZ
zWt<=7huc!@)_hXQQ(}+)6YSyYHZ)rW@xTumcz*aC{2P*}d|?s2+^x@wbu4&Ox*r*R
zPp9^^J7trGsl#!JJ4fYuR&r960><lL>ycnGHD52edejx4&>4Glr&5jRe7X#<gHe~6
zl)TUs_OsQ<y!8^S-u0jdmB*MAF|2ueFkfGNjcw}i<L{S`qj!_dXw9-Ntn9)z1kO2)
zm<9hNE&T*r_Dl$~`eKKdJ;MdZZoWioRcD#A=`Bh7^Ag!`_Kf@7xyAc7#$w)-ku;#w
zKdh28n$(>U-S8=5zaqTL9c?zU0}e@S`|hpK&2C|Lr@n;B%@l;*|7+4uV=``vJj&5i
zs@yEnNpwlB;&<aW+P=LKHhZsN<UWxriBhAjE;C?uJYGC&USVG2ECh#yQ)^5hDQBI)
zo0|h*pp%YQ<2`UzKY}0VsOJr36Ijf{n+V)nf*IXT^FxbP;dHYTz4*_C_bTZ^ICB7#
z2M$7P;wv(`aDxS2N=3+!0@>|T0Wzt12pcz8Nm?VkoI(9&k!N!U0s>!gLyK-~=)@}6
z6g)-Wg?)r;ESe8~EV}pFH?UQ3-pbAlH%S{ZoT+3E!`?Ycq6e^!S-J$uPV!r%enCYt
z>|C*lPsnK$I`qfXXJ;`=yF2;Di1*SiNzCKU3oQDhAg%oGFXZ37;DcVr!oa786xNR@
zwaXkdDe01@-a9@yEs5;a-XJ3X6@G>4P<VG4y4IH9<%1=No%@5YDrz**u5lE8ZwFL-
z@S%T0$Nt})Lly1f4r3u$?cV>$SK)zVQP%{~m;cV6?$)7_lTU^Ja~7Ptr$Ni^83KX|
z;j`@@zF61e$&E_dUapQgKW;)=Tt-^;@lf(ONPB;d!>RQnv3!8MU>thT)m00zJl`8*
zbj2=p-(%)4`e0jnjlh9#@%&+S2s(Q%lPLe|l3Cs|5Q}C$g~8)fG%9Tg_v*c!tIk=2
z;|o(*a9R{zPL2_5;bG`G(MfPnTTyPd9pQA7Y<C_c+o~`sxc5(19sinzEi53}E+t4+
zRi!gVi|pz4Ol<XhN!#MWGRz+J;X7w#W0BV};VI35^W9Up=xI-WbRR3rn^}IECybH@
zqM~vkWZiRNCNaldpZ)CAh$8$RwgUaTMsZ8Oji?zLht!nkw9%+1a~Awrwc$miqG(E=
zd@f>)a(Ahp%cP9yWiJs~k__J|J7rCE+SnK7jSbN|QE+LZq}PH2*lRit3bwH@LXwH|
zz+kKstP6{C7iC^~AMt+q8CK&rg{o>lWBHNUkQdI!&sm-@v^L{f;@s~4?YWqd-ryHU
z?WK7(|FHaKrpV)p9KPWtN<H&|S1O-CkMm>cwZl;|sGo}C(w@@M>kDN*13&Ybryh{^
z)KH1y>#fuj<pVB@CZj9UX!o24xc)H^2^Y8V3pmBoMb7ci1P_+b_YI|e6oY!XUle7r
zp57G<Lf@++ao4?^?T9m^HpMRFHE=dHsP`myr#c+Z?TjCbd+}RCTu_nF4t@7>luS{9
zyXG6z-u9t)&$}RXcM4RK+;L=ePpQ5^3zur^@zkvx-h*e;nQx7Fb={pjtRB<v#2oan
zZHN5ADBKN6kkwnqqO@`_SBzT37Vi|Ewu^_U<oO=dCa++lmL8=OGpgAF|9F#}R+&UL
zwi;2t_EOjGgs&a}xOK3Wn>H@Zuy|$#t68eFyf%uOmoK3a%^xLcfgZFMeKEFw7{dBA
zpi^BOu<j#bKlGPQJ={Sb@BEO2Ctg8BKSF}>Hx@c~7<FDZ65X_;Soo8fXczgGcfse#
z{I(y@yQNB}ybNfe`c(9f^G8qN3RP85C*|gsc&+(P_IC9{kkuEy_-7XT5}bi+Hid9~
z@)WZiT&Te_4(r0QP*EO8AuUR1$e0F?F^9S3=eu~AcAW0NO@w~nKG<ihqE&%+Y4Mre
zSa9VuY~xmAzrr1SWC>KvyR+qP1&pV=a|)C&-J6YUuf_y2iMGbLDO0e@a2Z`L=>q5S
zJpTMrPdE<g4p*b+vVpBpP|;Swy9b_VpS}#+WzL9BdJAZbXBJ}L7xdPFjd*+x65k%=
zbVS?}Li3Tg_#BPaQ;;U_sYF2Kf4KODP&`8cqeApCOLsoP!d0-f)Dm_PzG(RKRbu(%
z8)Dzwhr{Mgc$p>q0acaw(keXd?q&iGTgab}$^#z1rr1b(I-Mbhvlk=q^yUMqii?K6
z(lSYUwFiov#2$5UD6Y@XptYOZ$=dE9GuAQ3fVEk&F}t>);IAckqfDHm27KToW!l&H
zn8(?EqKQ97A;Pr+4>xDPr)V42HPu4hsU9<Z{HZV^9EuCCQRAl=UTaz{vm1Gu?KX;I
zz1kkp$oYG?pREfO+8JX-_!wrjK1b}m6L7wlaKfGxqlMB|I^RB+OedYfmO)*mJGX|>
zl*MvfeeVFu%5EX;&1I6IUlnk*FA&=2x8z_~(V+<YiuA@g+|(|Xj0O}UNUxGyUaev)
zw<~9!XrG4rA)r@Lbx=OBh}8DJ;&ESVC{0s%9u2xk%U+3$@1H{m@UEv_pC03mtuOA@
zdf?ovFjjN#38q-fLq~YazIadJ;UCXp%;6~f`m-Kgw47*Pcsag`eSV^u7A)HP(vKx?
z5V_6`%eJp31G(RlZ})TI^g%&7MlS^OiWkAf{s20~N3!6F=a80>Pw!2=sNXtGjIdB4
z);ErH^WyPyU>F~Av<K>|E<^J}F)g*8hqynv(EI2_^P0|)yR!os9T>sKj=M^&-_rT!
z%s83S*CH0=p2wV?1)<tzCCYAHp$HdWrl?>`KQ|=MrttoF!}4hEzZ2xzB*WEJsr<5=
zfoxb;k{#MzfL$YUknFM<8a@x;(s31aZvP|$z6v*J>05Mqw3!wi^M;+jDa~6OCyB|a
z=9_%B(>Ci2mY6=B8n1g|cX>HoTjh$BJqg@k;3XQ>;f#-;g2_EdALACh#Gbf0ylUf5
zel%esY$KlWgslN|>YfJGx=)5v0-z|en>*HBpszZPSeagm8Eco3o!EWr{4fMW_Hv1l
z2d?k+q<daoSQ{$iZ$<9({31)J-r7MMOZwyC4s&9guh99klW1*hDze(_P0ZeLNnJ%D
zR+Xi(uD@&O?7^YjD1JJZ`<g_Bf4;NMZ$(eBJb-0cOhWiN2Xucc?zugGvS!hBG5&g&
z@n<I3=K6&dJ6@GojvEKR)-}v)hYt$2O@pc1cv<G^m5^ynz_*TYgsSc*RihbD*%U{C
zQ*U6w>MJtuKW}AcQ(v%GJ!YVw#~G8hx4G!Nz5^3pj-kNC_SoZb6*28cuzf-)?86iU
zNA@@V)E7hUfiJ%Q9#4C{e32>G3Q9{Ki_Ay@KF^*DRo%0^byfvU{O!sz&4)6R*e#T<
z7l}JF)dW-E20L46ORW_b;W?mI<gE*tQ>-hUTc-=><l8K&<OJUl6pTUgS%M|-fDMb-
zjM?3Op?G8wdM?sLY{z@r+oh|t?6Emzz1+-$Dksyh9o@*Qax~Z4yBe?hUKhJE(d*!!
zsCZc@hW3i%Dw8$&pz;8$^xc!OFK&`VU&{@Pjtr+HRaZESN=EDD4QSn;z%*tJAshd@
zOlD{zxw2SJs%kfdqOJ&Ta@%!spL>hvB#Ar0nN4(PMiIZhX9u;3C#HSEER1x|rhe5Y
zX-}d%UW`r_owQq$WdnEdoG=Z1`q6|Jw(oFa$wI3C!^rfV71TnK>GV~if;)5JJ)j@T
zr*@^pVtbKS@<Qgs=ge&IY(S+P8~R@)|2@<3x6^o&aKYRTIVIe&I-R9su1=!~RRR3g
ztvr6}crrpQx=3vY++wLewo#88>!I6kBpY(K+T?=8CG`Cynzt&J*pQ?`r6MQ$@Yyg_
zlwGG~nqerMTStM@Q(4$hH-2`toHS#P7fzOZ!it58)GNG~bdt|QvOJK8sR~v2I^hg1
z4?n_2cch`Q+=U%l)Q{2(u3(d$JZU=1XFi>JlSb-<ioJ~<N`vQ8)O-n%wwO`x%Hh3+
zj;AqOTLtG(jpuw9J=KjXX{P2*aUV@)b<d<CUGN+|wD!P!(OLYe7x`}$XFBa%0-X{S
z1S!-*DX4R1W^pbpd$9|1OWesh*qpe2GgUU3W7F_+46|xUTNa0j{p`3+aymLqeL;3t
zhf%y|5XP^XgU<=EsEEp^U9zt@yXH3)H%`ZcNq0zj#(y;FX$;ajPT=nmT{3;UolhMl
zhgAjgnet{~wB|<wjR}{_%q;T2>?&LIdKgSCy2}u%myQ)3m5Bcxj?44f;Mfw7;o4&W
zT`f$9+3gVY-4urHf-A9H>P|`*t6_JzFFCGC&hT!!jqZz`v8qoYcAj{K5!2V=)Av9O
zD9&YDME`heV+{GdS%cCp{@8LwuxB#DF~%iX_FhgKYm&y3ee+)&*Ls0@1AAhHU<@=V
zX~0b?p`z7uV43lj>)rgwCN45ZM06msW}L{7eIEhCPm#<b>pfMh=^%^ym*Bq9jLTjM
zcImWB^!(yEa<rIBKd$wl)BE#bHGhLlW9~*QZ#)g{`X+vTN+qLji`kl9Qf9l(g0A-u
z!Yrpi9{<>jMy!~H{X+`iT9<%#U;ELhgVlV{gS`|oDowJqQja1x>!8Kx0=&$g%glR=
z-CJmrMA_>R^aV?6$iY3dzx@pRviT!(zj$6YFkg?l=?li^@&M8+{svpy&Y1?Yb!nW<
zOSoq&mGo}*f~z=NR;?H$J3ZtzZh!TbXw|x5!oM9*TBifc2Y-0opih#9m!jL#WGy;8
zyO`X%7c{tN4Q2IqM|aI?iSK~h*zJ}=>tDFhj5*pgU|tOmwSC7mbw8lUbPp*rL%e+P
zk%znWA;$?facPM&dlmA8y%js?-rqae$o@_=(#Vs0gpFqpm;YdL<L&8b_$r<~c{)13
z2*S(bTgdlM8WwHYhg;g&XwizHku&mT7Oq7wKC*<aRek5x>N5EHZ9&H!6)JMOgKPVq
zl5^@>KCnO$S@nk~*rA4HMn>XhY@OiuI3VEP3O@VCF0#BHi}K%M2B^?UdMMqR{0iQi
zxEr-I*>GLVT`-w9md|1d&!g#Bh$r(^Nk)U=YZln*&Ku@dLSv5=K3lr8OO@Z@G1ZDv
z224ivBst2tx)k3xm(s>`dD^zajK>YrMXx(v82RE38~E}h%K9x7-N~`6@p2>9T=s_L
z>nHei;tXu9*5b#yGHk#j*{mVA=;j0k=q1>Y3X`J+=Spy-P!ny_+hrqnFqv0OFdbEX
z3Rj00Bpb34Rk3?v_hSz3S^LtTPeW+un2R{sNs~=`bQn*D#L(4+VNet9xR6C&Jo73?
z*ZKE2s&d&M@mYM6%SHU2S)h;SNcKR1{vBwB)}<iu?jk=kX|&1f`+;mo4^4?8d&<^&
zo`Gqf8m1Wi5vA`VaLws0dZ}zC%Yu6L&b<XOT1v?JBrm<-e;R)G-Er;GbtsPQF4YVc
z9uJF?^l9mjj7YsDeEt6OlK$;_@JU>Oud(A$aLt|4z6XJh+#t2{2Ow8ufK^gmIyNR5
zVJr2i>||fuR*9gM8?KAr-3l%@@v~%W_GN5Vd_zAo>e=@=4$BE962~9)Xv~%1lE*i^
zdDlc=P6lK0%0hPIbrzm)PQv6hPGtAIgUP=YXa2{*Oi#NI6B>-^^Ei<KcL@=GIt?T*
z$)ZPh|DyvdL8QBu`{;#9KJBpLMt@>Zl(LkY8B3^4T7jZdBFi!RCwfH$pz*{xS+8zx
z{NH&wSS=KHOkZ!oqOHN{Ub<{<WE<1%){Jtg2372Qfsb9=&~@BwYPQ|Q%0K(C{&ii+
z=7lQM!)75>ZVs70a^Q=H=P|Wz`t<l<B}~!`@u<#~@=rg)d6jzXsa9j%ulB*J4VPgc
z-GIpBVF<15D(O{Ik6z2=DAa<}n~h!=x^OwG(_=UlatRG@pFl^P>7im5X>ai!lTDQ}
z)27`)n!P9b_9^5Mp-R&HvuYw=y&bg+H)mMSNuw)GBD0qF4;wn4Cbu(}>5q2`oX1_F
z{E1~OC;m6<93BUys&Tx$=owOclHeTfgY|X$C2bqmknZ|6bV-Yqq)oj=DTX2J+A9M}
zZ4E%>6FK_#ITz*AdP~(VPoYrLO^CJbL1711AXn?Q<QX$1tH^ey`K?|uTDzHfElicQ
z`(!fb=u5_R=9QS9dx@JJa-hJnsXR*BpT)GLqg>|=-%=+*l2tIQLxOqNDW~}B*BUTi
z(u41r(F2zqpOMnYv;2JRB8gmuc$awog_3vA<dS$LY3}=XNLEv$*<HNo_RD(I<U8>@
z9`2a&s-30G`z={>utoCmx$uwmj=-VwcQN~l99?(i^k>u=NHmn8|4?d@GCcvx6Js%9
z<R0?&2qb3>cTB9w<_@__Y23OGD7cbpVle0>H+t8|<b3v`D&iKNtvStX%O;Wh?KoUX
zPo$T9PgAX;V3^jHl5~3tZ5fk5J5}VQfjJ8J`)4O#bjyQm%AzQF+%`zciqYJC8Q%Xh
zL{{C1bi2i;x%%>56rxd0W}By>ecD|(oJ?kC26UyS!P@BM8-Xmv&-CB>N2Duu6Vv4<
z(TCF>Sg9}0gdfjX-iHc)$MHPon*_u7O)=#Sxrb9}C%JEqC*=EAA!$K9k1RaJt)5K6
z!~1ind2I}}92~}<k5r)BPWy3TiC`?g6WQVY{qW(6=%WmOgzEmHzuNyKPgF`tm)kg6
zGVFGsNz6W5XiW>}D*Pk*$SbhX%C($te1m4uXqs`n6t{wR(&%#qs4rNL(t$H*uzEFR
zAG<~JN7f+D;IB+`|6g`?_ZV2N$Ru;o@jU)6k}EF%N<AGaaLA;VI<@3-#f;fx(oG3x
zBffIA>`3;*`YXNnQ^2!>=}5~G2G^ah*j2r=h&I2^y&j)|d?$UT_gaNJ9p1;L@BRv{
zEuCpi!%9r9y+=l8o8ZjOu(5ff2a+$Ny@z|_o=acaVf>qB&8(6XcT6Ll9aACaVt_N>
zA~9NV6_5NaIyWPh^Mf~aXvl_e1XVl`jJz5wjh@9ToyKA4dLv3)Rgp3ETZ!OI<RI#b
z1SKgY+-MxfrAtY8|5fyc4}Rh2ZHn<@<ZM~@+D&xSYdPO~%!?bI{6`<yTO51%n|<mk
zvYp+>vJr{v=xK`~`n#Q>oJYYha6ZE|zrK^GxP@~2>go9Y<vslKjbxSu;pnBRPsYD&
zNMo-nm&~@n%^#VPe$o4>)_hXNCg+}+3a$TS8sq-5YKvIPDj5#B1DZ_R-i;@VUJySs
z8*IFuU@@OY-_$+q-P5JqaG)ZU3g6^px1Sh4#Gk#ns7?|6@8Lc>iu3zYv1FVs(Vb*;
zSBa*=J^GZBZbt`PQlKoq7wav2>842p!e3m%aMjmne>M{Gin)|tD~HSPzi}}=$GK`f
zrq+L~<Yd7Y`n^Vn3_VlOFeOYTn0Odi;7xkZ-0A(ka5Nt}NSzzk$jmN>kx64QX?K|p
zZL(u0#u`z{I5`~t8-*LX3f%7e1m;?Bj~C83D9(fu{MEh8;z9y=ry65u4X+>zvP6AG
z3)}JHv#g{3DwO6E_8d`>whu6)FU7)r@#_K7Cp>{k+fnv1Xg@Ogbt5$&k?F|&%iB^1
zbNPW)*f+a^n<v(o*alv~@M2%;FS0jk-&Cbv&Y6+Mo*$HXq8G{A1S3rBYKFzU%h)o8
zGehyd_W6S!_fdQ-DL-$IOYa(Z59w3Zy&#<_Zir)QiNX!wdkZ~N0ueCd8ebi=1s;FK
zQj=~j>U%!Lp(%3GkWmkLE<a34d65`7Xgkev3}D7%X41Vk+RUeMAU03Bhd0OO;OJ{_
zs4f%C?+72+<kDMw_<$yCvi0WCyUazF^8(pbn^N{)DN3VzN)N><<JLwc7>S;8zwlSA
zf2bQ@95oKvW9L!zi)`3A_;G2&HTHVZV-}SWAW43^hEA8-h<CZ^F!I^Y9nY?Y<+q8V
zTOG)bTrbD&Wikmrl*M(s_mB>DDWXd8Tw62P50xwD(qCOa$^E}Mc;RJ;z(7A_B>7O#
z+g6jxC34cp*G~9!+nO%!F`~G|*Vr!ub<u-a!kn_>(KEk}7R-2$NA|@~tq~ld-7Xmm
z9*n})%k!AugHF;l8R>k-08`%dV+#fT&Z3>yzcbg-F7V72EQz&lEGX5DPRUK=S$n!=
z7K+*V^oobbL>_n9Y6!m#*QoPwCv?&{O`ZRoV;QZtkZQ9Rr4@zz;k;{9*CKk4DIsWk
zznh-C^M;<bfo!nqB3!%sj!$Dl=o5D)KlQ7Ueyv~l>hb_?YZ{6!g4bknGYy_5PnpHt
zVEEg#@cGJV7(2@h<EP$bdAVMQ5Iy9I2UU0`Ivxr3zcFsc9n3#7np6UQaO(;&AN+KM
zo|~QFZ=W2+g&9$heKz9<-VMisY5u5uT_#&@-8*x}hKo=TT!=*$%iumf3}50yS&ZLg
z>~4_Y?EXw_x9$nQ)4t-_69bc3^5pYHjc$}o0S(whOZ~T^x!Zq;*%Zj<E3C!rNmgjj
z+)d3FYuEtcN7Sj<&ki_0fX&YHFzvsC{70KWTsCOTs~1>!Es;h^Vo<&REczeel$7)e
zmLH{(<0i4R_-rosoYqC^ba*%pH5Tz#qYgsLr-5o7cA*K@{*>~}n&f`9LE@T^$=xoq
zm&Y{eP0f5tOSfi?Pn9Lz!V>6HYd$*0beC?^{3{&w6_V&oO>B*|gkHY6U}@e$*2d!)
zTgEYWax-&Zw~uZfKT3<?h9Gu|T9##SjbrL;X}x^rWBY7`E*;B{A8=uR^X0H@xiV$R
zs*Ddr4yOEPYsggOWwaymC4Hu!por#WkUSnvn~v4stEUbxtU634&-P%^DJvKp=mhVe
zo^-<MCe3-1g=zzHT=KKR;9@&=<xRCj>qZNVGHmHnmxU6mf&{piX7UHu&A8&vB795U
zfzmO#H1SRXI~=r!c4SQ#UCz%4@}ERIo)zPU(OwjIl_R!p7x}(FKn|7J=%pmbeTRnO
zypB8um!^>a5eY4LsV-?d>V=GvQ)r2XH>rkpla92xDSRm6{cz$d+Ss@f*Saoao0bRi
zptJ4hug)Z<O5QYn`f@g=K^^bgD&gN@%7UAopu`{*n-UyhKl3~<(R+arV<q^p_ck`I
zuZ59B#>IS^YbS(a*N-nyx_X9sdHlqu2Lb%MA>po9Cbs$c(Av27lJwZ_Qtx^u>w7DV
zp9-qs-S^LB27@f==Z8gXV2B((_q&g+CJmVH%c#qOLRtE`D&C>>l@HAS1lPeUCF^Q#
zV9M8cnm;WbvUjJ*{r(-4{Ou5|ZO*sqh?)F}E}1dg<8k@LV;tC|F4*Qbxn<`??6b%#
z7CdcXg$aS|N0$<G1PDgllRtcL#&ayan=U$ny{RsECUXo*;p5$Z!~RBucuw8MkKSE*
z?(2ula+W#tB`s{PaB&n}KgZIJl#*iVd)m0w3#M^(^kJCbJD$73WAf*-MDrTDlH5oY
zvB^|CIL1Ud!V}9g5250y$igjs#IsMSp<FwidAiR;Xqz!@xTZqulKjc$j_^e;QNpSr
z5;AEeq<6hYMctQ@)etE^DmqB@_q$5_4YY#WsU4^k*}Q@MFQX+iTzscGVey%88ibwn
z-v1Ok`KJOtR@2#@!`3+J+Rio{ZeaF26`s$hU{kOCvd^0F7;<zt+m)1!ZZoq*v?>(U
zBmN-ut0ndguI5L35YA>4Vr-ZPnl+b8oK`Ae<AW<$D`wRfXg~Vwsl#6}kl%8H;YX)X
zgteXLs`?3h?w-kX+~)<2+}_N)wC|D0uThddt~$z+pRR>L?n<Z}RS@hfZ_*!rnW>J?
zVwyiAn8u7_Y-e#OyR0>xzKBkW+2TaOCkf+@W9K06=nUB2I!M;Zsci0_!7vifp{e$>
z;Lx=Qk+t!drYmwT$L#2KtOqyTATkeqhhVN!2{s-n#v8|2x})gEV`?t5utL#;zan_G
zW|oN4C}yd~eo*{&5Hl5WaA46d+5N1ILbUBikrgr6@G*sy*B*oUj~o-@$#?1M=f&hS
z$dwuOy_k`6p*QNc#$#MS3b%9*fWfX{o>}3=i%k|wM0b{#pQ*yx@LKFVdKvnD323-{
zPIg2qo?N}|i}y({7}Vb-%Lu}7rO7O<^F>KpTmTQBlM5!e5{YLf@}a#FXo44^`qLZ|
z-%lLc@}5e%+3=LGTd-ZyUG}TL3uW#-L^;RP_^Q{3(0SJz*=^ZnsI>UA`{K@(P|}ai
z6#nL`KN-Pv^m!6bMc5YPu{O05+23U^_>P0eXx05>COz^6n+|rts+oWJo&A4#w>))3
zE?bTC<tFGmzZF@p2hyek1<cFrhvc2J=qgOlg^sQ=H9b6w=j|hzrREIGFMlrPmP#2P
z+ccnXKoheiVbn6BTH^g>rDRxm8gdeMNZNmid!dyxnXR72AO1<_75`m@(Hbwh(fA)d
zwyhwqrtT(kBi6!p@I`JFJpl6JyBXzFM?-AqNw%dq;QK309(X!jHcZ=>S!(F>lXos6
zBr}^Yy7Wf$JuD>3+D;OivIA&8Ka}LZZp4D-Sek!-BW+u^Q09NG2iDGeLv70*Q><JS
zQy<#|9nBqdAVV<qzxAT%Lw@X##xNdH{u7~oF|^A@baw6?rM+ca$@@zpn=mU6f!F5?
zPt7iFrgDjwRZU`*y_0D{-Z|>l=@)VFxgYRY!v8zpge&Wp(#@cIbVOY^sy?K{@vJ}e
zi|er4?ho~xnIJf2!|BnvjZ|Kf3@wH6q}+ZEPtPADtJxOpc99pG^>z!a1qWio;&3|l
z`z8zNHd1!IVIDLZ<%A>M5?}p&5O&`M1-%<F<k}Ou`M@2j!Z(or=9+B3M<QO1@I}et
zCp4+mkF*;PNOlHorEVR5^i}H@d$@WUa?|I*dX^9K6<G`I@m0)sX$setCbCDY8_s>m
zV}I9KVf4DKI66EWNxlmu-l5)X#r%onDYuL;ZJg||^JVy%)iN`SYj8B{hV6nMI86Eo
zm!-e?p}9f0B=W^6aZP;m7aOeJR)iF_$r(G8Z(@I7D*CMLp!LK2ShU<qiQUurxNbO`
z=iM`-0+mkEqdCv%;9O1e+8;|6qgvRt_m^nAVHJ8z$ic;T-Ldn>AgW6_%uMgL(8o{P
z;Tp4>O#5m=Z)h`1p8k+q28oAD_eI?9k1N?Oy+uPVdLw7YTWrY@UAUyKc;xdH|F%u1
zvJO@5@#Q3&XtDxVEaTyN!JAI~(_x=JOy)-xm*HyvtH?5QqKk_{NVm~|yby+%?N=!J
zv|7gW;|W+cC`B^+UorPGzGmW<dRi7ZQX9&bqFB()Ao6XI(A3J|sQb^D8gBQ<tkJtb
zA6|#CQG?ZK@r$eU;N%Qgs$bwX%Z1x~n3l-f+`uUnSI9qcktoLHFxURGX;`apxQ8oJ
zh+85QTLq7%#)X&V4wN~)xk%pIu1n?(pTHa3f?)O2MR+C^xavA(R@1S7E<|2{WZ_2|
zxmOKMi3Vi<=7{LCx}vEh1Z6%d6lmbjWA%pP&Uk%3Y?30=d}Rmowk;Bq-9ykTGL`0j
z?2ru0zK`cSgLtC47e7#|fv%y^aGu_3VrRV!4tMkUK?5&V-ou^Q)%Rdo;(Pb)nlFy>
z;nXe6Q_OjaAd7k={#awhqs`FWm4HdPr7S>wCJ$+ugID|A@aNxStXS_1O>-Zpx@yx=
zE5;m)J0!n~Be4DcewYhah}*U#K6>|4iPrV|vO{)9cyUOh%sxL23&UICUUEcoSg{j|
z?}V|SAQw1JtbyauCiXGT7rV!N;A65fSl)O!>2i;Wa4Vbxr#)BUum6S&>^IXen~$;`
z&-Bsjq+Eu=D57f~BJbc_PJF;?_T=*);bttuEODpVvCy3oW6N09NOveYz5!JfAh@y*
zuJ`LsEB97YXyPLz^n4G#YBjyOcbA@?8^{j0W;53wcle?yaZtaIOs%7@$U=vlktltN
z!O?(N(%USY#UpP+cg0ekCNh<-C;h3<@C;c>a|yhwwW-sVK)PFfon@JhlyzQ{A~V`G
zfT|=(_*}ID52`bf=r)3Ol)Ph~itkY5@4?KllN;$pkHpI{qp8={g(eR^hG9Y4RsQ}!
zIAX)*Lc51L-Tqbtog@_=-ou<q7Q~?8={&Tk+=gLNEgz!tjpdYl!RZM_l9Y+*eAG^V
z<o)h0bzT&NCez+zYbba?z0COMe`3yXG7cA`Pt&&Cr|hHkLu%eLjB7reNrABAw!Mep
z=F$Ifz4HrN{$D#Aa=!}a#STq9VK*A&V@U7t5_pfCiqxqiY2&WZWNPh`K6>+frYB2B
z?TbuIT{D#LTC0Uy_vX;#Gi&&yo${G?w#!kGemKn6H{+XmPw5q#E6hgpvEEl@V&#k+
zt{^zgjpF^%xb8oA9Oux#C=Y|78EBVV1^vU`Xc&HhM)hhV`P{Q;P>Q6SkYRlGz|q{j
zhb}gHg!1#*&g6CGB)OmJLX%b`V&AW!d=r99>=(sIoHos5hM(Mc<^E@q&cg4Wzk7>J
z^;%zS+ZHUc8Mh=O9G^leH-T1|oMm&jIFg-wI4|E9D;$TPSkluP$(pX~VXE3W<MX9L
z$-2u9858a}k<!m}8h>Rn97A^SmfjW6@R&%)cKXwlw_<<foJ-f;_p%F@9C-Wm(Nr`k
z2b&|sz4qBwa@n1Ssfs=DD)bE!KaS$}cdSLPdYND*Dv+{pGmSEhN8*c}V0l+*cJ*nj
zG`h*2(_oscFbw-*7sDlP3a$vh*OM2P$P>M#xb@X2bLot+?dR$7nCr~!y@Ir-><%Sb
zOHf>A1NF`2ETg>)Q!+NNgDTm~Y|l%~UZ9QOGd}E^$u7FIITzk}Ke%1%Zg$YU9Wgz9
zsO+yjUdguOh~PnJ#@F-t<NYCQ)7V?=gIfnCa-Em|v8?y$vX)nAY|mp~Ed25UciBeR
zn`LoH=t>k@DQ9}6zvEqYuHrLg!hLgDmu&7;VaJfpWd9_EJ=z(9tzKW4Za;Y_xMm{W
zGXeem>tMet9<!dNpup7`g>4oTdH(`dC>YV_uODC}s=Fgbu0`j@lW-Tgp$qH%VOg(7
zWAB~gqn-`I6|I}hX|5sr+0qwpUw+{w3#PI=cA^`*;(*|<9-&QbclgV+uH>UC=IHB}
z(1b{i6=&}7fUUmRJ6Q#%M-FGlRkNX$>BZ40MV9BR#&-7qfx*$*e5C&wGL<=|UuYVH
z&aROd!Y4pc{6B3~Hjqub4YsJ~vfcsXVcnb{8(!qeKP@^A)$DVmFW(nwW6fm~-yRXY
z{0!1plQFG2xy;MaHMFgF20Qdjz}ft|(62t%d0y{V68(%6tZ)g2#u_=%>mE)m8~w=m
zK?(I#Q)2;2N2sr0v`a$Vuq*Q%A0T>#di(C;)tCrm&-un>;vP5RS^(@v+(9MfV0rZ<
znq}Ufj+zA0g11>r`7OuWak?lRtth>w6b<83{h%^MPG&l*v+!ugVdaVIOy{CEpV5%a
zUCu8=ic>2&FD=5^OU5+sR3xvR{gbIhFQkLdy5f+V3I5D>q6^OgWhL)!B3Cy^@JGbI
zb=&|%jEzCrghf=4HqONO+BiHfi^t~dK3FZ@P4r$r5}qbSnR|3TYhU#b%CEGjVu?FR
zwqC}(VZwR3x{02}-$lvsCbmXPHS<^Vb|yPAm^-U1X68z3*%ratP@fZkoKOX_Ry_j6
zo6blN@gh(2k#yhVd&V2v2!3=z2rY{X#rKR_{P>W{v<&sIXlOi>^j^iME5}2#-+c6b
zu$4Ue5M}p?rw#Gv>5DiYTkmXVw+?-in4B(`rT!O47F90z`mBOI87C$0`9V-L%f_m@
zd}iRd337eqjZ>Cr(KPvTs(cfMYdxFUZLyC}F?@`5PY=Onh8!~!?-%ON;<3D~h?ZP$
zrWw;>@nGvJ<iu@<*VJoVzmpsD{;ElVioVdSe~OggH*hcXq{r>e+^?cU^fIm?Ep8y)
z_dS4#S)JIFxxdLayoT<~OC=qX60+Gy)HfgoJ&o^Cz19`7f3T0udUqM4TSs$OZB?m^
zIGta#7DetgFgPamWZ%y7e>1fO&*UCoG-4+zO5GVH_NUYpN0AiuNn)wE1IO)d@y75y
zB)jRu95fE$xp@&ydH5bR!*Y-{&fMhEu6%w--k;gqCi43D6(k$&DcSdKA+47_L2~DK
zHZP-w`9+nY!>EfiD>Vrl9OGDoeI=|`ma)lNel)9hFI-Ulk9=ZuX-vUKT)%e{k!up6
zHNyx#-S4oHbBfg2b+9Dq=OQ{dOI{kdKta0nr3On2s;0~5Ll7_CT@(BL<E3#I=;x>+
z8gVocT}GJV^R|m{P4uDP%H2^Nx}H=X$FsI4#&8$emI{Xqwm;Pkfjc8)SI3^D<UcN`
z>^GR(eD;;BZVyD{ff`t4pF~8}Uta9HkcYnb0V~ISA|H7Q@&jf-^-2q$a6Xw#ZkA9=
z-6Ogi^BwAnlQ1p*G*+30A=2kP-*kTlb9tuBJ{$gJ-^ZnkyP+nZGB*(a(iAccMEb;d
zT_T+B-6I`?AQ(D@;a94B=IiGlXyhAH_Nc!L5^pTWTsJ%1UgE}-hq<CnZY3!n>4RHY
zouz-oG~0IDY%CM5(3aKCH0H<|SR8!Gu8yj~r;2REs}x{S$VI5zuM?SfTO1O3)BLA_
za2;!d@dm!|l__OTKRc0G)+tI`TtBnkBQjXK_&ej<nsK149JZ!iDBxfPj_>ng`iE2b
z)1)|b=yjrjUPDBe$eyVi6v#dw)TX>A?zCO(zS>UwM-mTD<`>$9%u-(nj&}{IC*Q-n
zg)+XSyYTjI_QY|&o!sk`8FU4!bW6rH?yu8}oToU*G*%|biXYyC$B=n=)1xmx&`=_H
zl?&n6=`OeU{e<P@dgDmaW1f9SFl<kH@xC4Tys7XJE(P91#r3PaxMMUgPR?SR*LI`Z
z`|r`13u@>QRmWyDw)0uvKk==H2V>kQRcvifC&z>vvZLRU=xT(R)8z?9xJ-e9`w2c!
z-7@2uqpWfF&<z||_!7%Ul_M@Nio9+;NMGZdOsaLiWykmoy1GK}{|in~m3%Obewn~b
zLl`{`75+W-bx2oy#RWbDj*}XBTcZRi85`Lo)k>b|7a)9MPaywX@P?mcv&5O>F!%9G
z)_1WPoRz9GGNQ}j``<xCkMD_LCiTV-9z-#_W7e476o>s<9#G42qy7=6B%#?C$h2A+
zJA2$jhK3bMF3rKqKS6k?x`kFZT)>^wvnVf3ft|}9GF9tjT;-O^j>v7KBlAD97sG|8
zSQOvtJ7<#3Y*XqP*OHOAC5rC0z2;6zNtm}vABS9)NdEpjL)Qv#@kgIEk$h$k3=M9;
zYi%}tukpmB&Z{tcrUi;NZNQ9%*SLFd3ET_ZVDIUJ+C?o?_b-5d^H%50{Y+)wKXY7E
z8-?0W<?#Q|!c@mK@U9k5dBC@2@Ei1-)?RcY^?3@?bGn21@k594;rCgbD^Nz*tA3=U
z7R;RXKftOtJ!q)7S0)J7MB2G?Fsn?)$VHF2agQ7t8nXbCdQ0e4w;I0hxFfA}$i{p5
z(u~dl@-)g&o+jMz;EL%#p$U=Qc9|gSd?A?Y46r4=u>CafksTHP7XX_Fp~$JTqPTrW
zU=`$q`Yyf2d;2-czwsG;N1H=F^CmWpe?)Wb#!^~O2^4bdgwHA*`lC}&cT)>Po~1%V
zrc7Ng6jN67Rs8>ap0%@(dF<3gzkiQtb(25(|NBXgx~|6diK}5>D%|!-uOU~s3WdKH
z$&SzNgiFsB<I$=^w6`vWeVxal=Wz<9@3LSqTrg|4Poj{0_UQFy5G{MZ2Fbx&khM4&
z9;==(wST@8tP@6GqxVBAJ|B?^8dy9_f!a+oAlW<x_Jh{(k8y9P!|N0Z`s`y@$cf3y
zcj0RILE*8uA@MU7d=xip1V!4SrMiK6PhUc=ZPU;?{{jUa-iv7GNZup3hC<I5f;qUO
zvw0Vub)_GvY$|5f78m)(A}7)lKTCYFjrM&IeuG(=s3#7M6Hj^9W;ZI^BZGb)MOI(c
z9h198)8j85__c07#@9`he2uo`9n;Ea)%iT8GEj8N&iA11k2gvT{0q79y{WigtIpJ}
z$I<7LElgp(2lS2>!1vlIGzH2Fud$EFlc<q9GoztAAG7-HVDmh`%W5Zg!Z72RFiP!-
zjz<gFP3^Iin!|)gtzB@GUShA2h2%zOFKCV_<bNA7aVlUld?J_P@bz}KYh$s*(P|In
z``tm&VmtiW)FRt)V-3%o??w8Sp=@!{DBkJGLsZPJf@iE8El#n)X#Jj0J?>0%I-+4Y
z#RtRBtz#^13I!Sg{g0XBU0FMudG-;vJlc!atP0|9g-hg@;$8Ut%E0F>^V#s-vtZR_
z3eKNciI1UzT|UYeKE+Qc=4~kRj5~%$j$tHG`N<Y1f$h@tq?d(8vaPE^QDYd0!>gy`
zi`q$)6nje+Y&glXyCjj%k^v&$avWc_-KSkA<PkLg9XlR<2HQ&x)7z=364!YHBxUKA
zvT3Wu9BH<2Ygal@L{vNTjVKlD<YaV|c9LrEeF*Q_KCGd}8lK<3u)4}TihLlvfb(2g
zAJ162czZnU{}xSWbVcX&n<k2OKA}OgbFunqGY{<=!8#39qbzZDf9c~v+uNqlzti&4
z+WFo{yETK(T)anakMGgo!23A$EeAst-;&*?DQK!4#Z-nx3C8{*yiX9lkE6X1xiOE@
zTUPSUr!GR{`x@%ELrEI6?=TE)rX%N`HM}C5F*#`~y0$i9X_u~$XQed!sTaIHCDHco
zYDjxL2Pxeg*st~y_O;V9nyCGi)t2~@<HkPHJMX@+KQ|U5(C7k}zxa>OtTU$co<WjD
ziwjx!phk@ODFd@8V!kI`C_CPY0@!L=zP%c<$HO6=C*0!wOrX-HL|xvz=jfPCThzKr
zrwzG7O|k`~(Q}S$MXWaB?qssgq5VXp{}RHQ50iuN+Wl2)<QsIm@HO@UFdXw1v7?V+
zqD2Ne9R_?0QmOX_SE$ODNk*OSfw*;R;n@(3(>ndI*2fcd-VY_y3c4U|`V|}++lf0H
zZ$`m!FANNCXD{Ez%Rc{5Vo$`(B5X?>#>771IbAK_#adD9-pXGM3&)0IZ+XNPk?s5v
zL6u<#K&kuKt|NEwBe^QWp6Bt~KRa>RJXLh~?;tUrLwB_a2%c&{X@lxb^ql*VeR%>c
z_E5uxH;-}r{X(dA9YT5;@f7!(gi|4lt+(0D6vqcc%TQnNIQP-<z(cI1Rht?M($L?Q
zle(EZtr#!zr+h4GMylb^)hL<EnM*V%;W=&i^oiV4Q!q0>pLT!AK;v39Hd`&07RVZ4
z`8ks>cF-VAtqhhtWixkJyc*hroief-O7ZY-G_=f(sMDa;SZt++DWw5$+B=+f*S+HM
z!I|8^*-<#j*FgJs6^oq{iS6^ECEk->Q}*@yXq?}jcI$Sf^(rk`cw{HCTJ>SB-W>;4
zT|tuMAQju`Bf4`IJ2@<o90q{0O2hDRiZ6Q~{Dfwj>|kRfuOs+%0Xx+@otmizHRAcR
z)o2+FKBf)L123p~Y6b#gj<HpP&PbGV?~>g_dj!qRgx%jJQk^Iquj@|Y;nhOQ{=JD7
zYwA(a$tawibD3g&#d)*$7qfd~0PT5Nk}>hdw98^D5}itU(DAY8<}wvSAIJ&+b5H1$
z9mFTYLZ)o!E%=c7cyD=M<N|fr%&T>1?Ev(it)}??2B_STjPzN%@V?Vm_A9{)4qc)l
z_a&eHYba%XU2E~D(uiJ9h`_TGVy{0y5xRGB#9XFx=3D!o(vI_|p`l>`1)cL`@FtTb
z$hkm4>=b_5i4ML*i+t`zihl5P+IOv*7TXmfyW|#kn)`r;CnO-YD390dSCEDtolEuK
zdr;EfBKB?3b9QTs_-@41(38l;GUq{mnBTO)vZHMh=(qb}M8-voo7hORg1lj^*ug#0
z<UyOQXqVRjdR?mpZ)*o6Wtri@{Pk$FGzN_+g3Z(f9vUIKRAriYbK@`6cf`mR%)i1E
zPxql1KUK6{nM(~H#qK!Ln3SHy%7UC_s1i)&{>Sb^J<$Na9rg-d@^)0ThTuodK)jWI
zj%u$8*@d|a;j%Q9)ebI_e65p{_FeN3aYOF2IF~5)<WQ7k!s2+QJJwflOLs!C;UA5f
zSAaQVCgASJC4BPCa#q`YkHk<W+{Oca5s;d~GHz|625OR6{Uc;uUzwpdqm%T%vWH~%
zU1XDD^C(NW^?t>)v&Fjm5ZSF3YVGThJ8CnPO)5poma}+!bT#sqiJkLAZ#FtkgPE?~
zCu#Yvjj|38+31=)*@u|ZJoAe+bUMXw>(1kFN%Jsw^s&H@1!1tsQj(ac5RX*1M*D?P
zP_-K*OD{Z1jdttVx1;AVzO{&L{MefgiM!^G@V64<iac!0jN%DbC(_H-B+Qy9qqm!H
z(2Z|FsQhlif_86%{)uMzC=8?0!()Uqy&LOZ>Be?>gi35HHloE*xa=EG(TftXFKE%h
zrn>P6OzKOBxXWh^oGO^ruPAZ&S6u&DO;>_`LUO2qRQ>;AP<}XiSlc5y`ZibVI}MLc
z45dAnQlQv70HyIi;r^ckx~NBy&U_bo-ftkrT+l$&@fP~eR3DxH6TFCkKQK;>XYU&g
z;JaBkwkrC={k}GJKePm%v!=mnT?yA58wHoM^3pzGfjBm4v~Z@+5ZOFmQriAeIASJ{
z&-5ASC$dS8e_FCN=VeU$y#%@$7ug5ZXvFm6vTtRzbaco(O0{ujO|2QwSv{I;>pn|D
z;?pqcLo0dR&%}?PPq52jw`|v!V4kx+6927ImWJ30rpHR<%<P8g)NrDnrN;eZAz#XA
zx#n(kmpeut(kL=&QYRPnXkI?^CAm7^rD>+8`KdcsXu#7LULto9s)nCfdBqgACMl4`
zi=2j^x!_T(xN6cj0LXTEOy(WAtgvh%g(Ot-#~Wr-QYpiMziPDMuo-UMOyPs3o|DBk
zZJ<9!{?H1wVaiF4G<A&NVXSat@`n#YwVY9ta2-0z?Zs)U3%ue;9*@pH#@h;&WC~dp
zm~U9acheR6XLx~kzm|h4GbTyN5-yX0vB)y;NH5J!z{`Nn(q6qzrt4-KFw@-w*n;_r
z%=(xCjcCe6Sn_yORnM0_b(u%~TXivD>Qww(Iu>3Ap3Ku_Jj<$DBswR;@p>~C8iPN{
zx?ZrNm*y^fZimRIHbqGIaesDIXEE)&bTA`syx57|ZNTZGt-u>!n$xPs+`sK+za7LK
zF=UmHdWInCejJN7Ihk?Q>N4p2Xym3i!yrT(b=LW~<Ti%ANZwB|_xfS9qZyqGPU3b=
zDa?H4Z@y;dQasDmgxR_z78mm$)xXKZgy9T0@{q0Z$!CLe8fo2vPI&oQQF>tFN!ly2
z2g;(Sx4eEKoO6c2SA0siJ_!3JyHk0CvZOUVhaJsc%KrT@p^+MKB>y;(Z!@ar?%L~d
z(?gzQ3ui;=;Vpzl{gL_qnL~;O5peDmh{AgV(ecxe$(`zw>2<UNX+LBl6WGf8ExU!c
z)gk1St|Zm0dI-nn2HD7uk>pUL#g=Onvde)rq_Wc<<8@vmN~R)Rr{H74myKsF*GKaO
z+P$&d_&Qt;oJ48<6lB&*X>!yirj%a5W}M%E(12r*B--O;r!17M@W<8*xiomnAO3OV
zV6u@brT6Bx<RAf-8D6KsUMb>ElE(5Dj3uEGmqhQepl@{tVUW_D9^A?!6}77-H%;$R
z!DBO~?hp%C9dAfKw8CleL*`Kb5ILKiG0fmY`Ub6RIO-3kA#nw;IkW&((S~@wbThBh
zFk?o~E|T$WYwUBMLZe>9!K%@j{`wrjpUdt@$Q%#d8GYHIe!e_cAwzO_T_#04c;n}$
zcx>M}5HZ=#+}SBDqbAjh-(RIe%T?W=ZIb{y!A9C~^M8e%2{@Hs+xAuFe?>GaNt22q
zlGy9qq9{%NlBALbl`@n_gCvCviAbbGgd`M|*z2N^CM6Bhpiz@FN~PgjJ@5Oy$NN0r
z_df4^9QU#9>AIJ7-+S-1u616&a{!O4T|(~N_K}$+4O#u=$*pk*&rF|y8Igl&%8Rp9
zXu6OsNOwfe+8FGq@q_WMOd55i8+`9hhGqC<EKaWB-ZH@=nc0`AW*OtPdxlt7+l=`5
zM8qEbBq>c549?Zr<`ql!pw2(ee2sE0Ezdl_W&XSA?7Ml~?dls?F4-bWU8cr;hg8vo
z_pjkx5JOW=S<v4)EszXt!AzyuRCu6C;xd06RD<%^8XI{W+VhdEE)2oK0qa?v;S0R(
zwVHzMa!U7%YhUh{=qz%*pP=}wh@G2ogg$TTEUjy6;BR_a$&z2+6!n)nG<?PrdQ@A)
zZk+E&g+}|C|J0#~2^$23%fUD*O-ADSWvD9?EF;T#SU=2G;`I$=bp0e9-8Y}-Rqx`r
zhDF2MGJ)mG>PXddh9vgOLtJljMg4M5ma7vhJmk*e$JtVmn`q#j`d-HS(%VeC=OmhY
z)PXgfbR+4FNqmb$%$?ueWVxQvD4m>2&DCaXdEFT{^2idd-YXwZ$6vziuSuj(+JLrQ
zFC|aj$(769j^yj*L~ZQMI+*3}N1vhBp}3$2no@nJ<(@ki!Yr!qE-&4-BA%eB4V^hB
zB=0KTBfrrSx^FEo#KVZ%Jj>A5!-ArY>;SD_#I_H+$jz<Jq1AFXY`z=F#t(f(83)At
z<nU9L8I(gGC&^RSftutP7D!z?Yhz=b9e)v&g0xUK^y;w$uFBf9H}o{w*&JY#AI*gM
z;Ukj1qE4xOl;FMC>A@(lmEt28A@j{siN|P9wA-l9e@nNK^Ts0b7Hrl9Q#bI978%Tb
zo&y~A&7z6Mhw)17E}kvPLd37d*xPU%cSS$j`5a&&y>C-mLntQi@?wobci8=I8NZ#Z
zPiYVOO3ZXT;MIGnZ2QC^uymV>NVOo@yUQD>DYU(G%pemsVc{j&;++=-uk$5Sh&{(|
z>CeEgYt7um<r7}Gy+YNR{`76Q3F)peKvRbz3=Vz+gODmD-mk#PbKhuNKcXDrdv!H%
zAi1z~*~LwVk+fwAAJ$w((H6TfMcV?VI^igss!ykOyuoef8eXf|#(pX7p|SaO?4^1g
zEJIH4u7y>2tUC)O=QS|z@mR`#o5eNvcBk$|;WVyw2RtH#hTgR=^*MD%XrhkryXQmT
zbvzWy*6Ei$O3bl%Ja;s`sB4sb{h@=x>)%lS_ja1tB@H_9XUyH!cnW@B3$*{9$WoJf
zV%!QhTC{%$U9F5E{}q|Me}^olGuDy)Rk;mH{$dW85XjwfBFJ_;P*FdJx-VCecCC>w
zFTL}dm2d4x1}n4R`o#bRE$<mF7^CONI?1b@nN(Jihr%nd>{$L_G~27-+WpVm@68-q
z5w{ttW14yQw^7i3E#nn2bv&R(ovhsM%jB27lh`(lr|tYP5_pa5M5^$T{WOa{M9WLB
zsZ3xx+I~E(pC%h|zQST!_$ho%cHx$n+xV=Lf0MzQF)->1ZW5=@wp$8c;_EeFmrK~{
zu}-MH@P>OBKEuMjV(nXK^PT(TvqZlPKHzI7IO&MGTBV=h-=9JG5~B;&GI(7WOkqkT
zJSDlBpgocL{*uz?<|^2q^A|bcp0cglGm)Fqp6|JFA0hkCaI0Aj=#=*kS-Xqiwu(tS
zYxm$~*L*Bl+)H-WwOmr!d;!01hF}dTQI{$G5V@d`cHjS-+{%Jj;Eg4CB6#+r4s4OU
z&=TKgb3CAzTLVeD6KdQNL>*Ba<R`^TER3pg^@B`eyT~5xJC#uFufNf)$(<Sx@1id{
zg`g4BQ1!+E?ujZ~_0tDg$azQRvMNLNICBj0c3!|Zb2;e<Q~C04p|ha4`#ZbiFdA<!
znnGRmoMiF`O}y29!jmTMl>Mlfjdk>!PG)XKkY^@71#3v>+_=GZosDLa@BY;5R04lv
zYs@q~wfTif)htkD75RNoW`%1a$)WN#f+DXYVDt_)L_v|<BFxH+qm6N~GLBhai6z5!
zwH6<)h0u_`3$d*2CO3RlK_)I`d~NhRG(FmlDUI<6OBD4F=R-;R&;{D(U&TDSo6rb%
zC%Tku#C9v4NB4<G>EynHvW2yYPzdaXlVkGPp?Fa@vEeqAj8S7&H?`2^qXy14uA=o$
zgJtzDhN%9XL7Th1ARRMptbOf`8%gopsAD};hKVuu!gds)6T+9pM$+wG`m!~@_Y1DF
zSijg`L=)vwaJ^(DTE65X%=kF4ejaUzD~ILnAldX$v9b-n_fxq=ciMj68N*t#xPGT&
zyktIE%$#jT2F9}lOX~rCmQUq_s-1BCX`IETdDd{KdIq)OE&RZy42j}{t&|_Lnu@pX
zMda|=T-jVG(Wk_da{72`YLZc6Y6b7%Fbm;7Mv1zZbEtGJpy<C2u-33(jFF9l()-_B
zu4{E!<k|+zbK1jlg*W%TzYgNaK0WT(%bQ+Y+$sET@_6t28W`N89lz3}jH)kZv84wG
zQ|Y-N<lJ%OA)=;#-TvKlSYb0=+vQ%iqOBU<2aoeT%NzJO<BLp%_Q*^`?%D2m7wX-4
zGbwa-=PRd|(M{7uxMip-YV<kmes3l{d%?)-xe6_D>C|g_5*~`{t7*kN*)6p^;d`)}
zYZ|AC+S*#27W3(hHsJ-RlO)kGx@KNHS`%wN{o*%~Ogls6$a3)(iRmzP>EgayxpDty
zp<`MEIXU;TZ_%f*TsoKenFsNo&yrciu0A3+)|=Nqaf9J#PQ4~}L_#+i)rTzO8ActX
zAw44Tb@c^Iy0M2DrAN||t<$)IP7D%!q7fXXg$JHdSp8U;Jj-U#GSlOvzxf95kywF{
zAPuTh4Mf;-k#k*E%!(sbaP+<z@9=jpGzzSc;}U|s%U`lc1qp`L<>Skuq13<2RY`I3
zE-qO&iA|(nxOyib==4_la``<=&z%mBf?K%gqz&`SS^R$e3ruWDq{vGi7~ZoBk){^!
zyR(R1HAWzC^gO{-dCkg&mOt#udGb!#LM3z4F=br=HaIrGWknn>eq4;90d|;si^%`=
zc4VKe;z7NPshGc`{x3oxxfo9x5l85E{{sFjAqgFSGFa&xLCozeJe!irKHmQ*Gn@Pg
ziC+83J6*;j8bx~pLdaXpMaOv_LpPaVZ(Wv9f$s}uryqxg&mw<YP|26rrc(65znICU
z?lk+Miqy*LBX`ISV=7B8@>`uu>E1P0IJ_;VhyB-Ls^KP-9K3|vcCn<}tWfTzZiG>H
z`_O?AeR+`%XWuqkiS_*_Sob-TrXP6@*DiGyPM=hy2Pd9`O0^-)K2rhP!r7EyV@|C-
z^HH~G52VgN*_GvMSk?9#7E)=BkAsqB*9#ePyHvxm<P8SCxhy&Q*D2~AW=;NIxa8AN
z4``@`(}vnS`V^Xox`;k>%(@+=`vjwHTO@PM>xQjMl#%7rS!(j_2;Z4|hTRTQLUF@(
zF*nI1zaAx$OL9RpydfCr-Vb2nB3J`kLs1bN$*em(A+;9~cw!%c6+P}#cduN!^v;uh
z)cGUG&J%0Q*RUp0x7uCgnm69L3?r`;I(X9v*(T{QRA|BEt)WcG$`c_k89i~W!~)F>
zGAk*gJRdE1-P=VEl1pjwhTV81VRYT$IFI~v756eVNp;w3+41uERJKd#pIYZK%XkNN
zBYFxv*OptjqzxtgbNiXrI7=+6R+0ug`k8O8JWAJ{PDuQ;wCQx<YBJSQ;?1iMLDJEL
z_T`kJvtuM`$IPZE`EIy(^)vHRuSHq04AxT#!3{&9PHC|85{%x4kNjqX3x?MXVH5W(
z#<Ng;SPDOtdA9O2ut5<+1pE77xD%^Szt79Jf5-gWorDL-LtZ|zo<8*qqVAu&@&xJ#
z{l~}H+WEt1VSEDA4H!iwdn4>(2@9~AiTsSsnA6stp1+MJef4O%GC2bm3};fj$~-=P
z-4<$VYn3DoDaN<7j}pZVA?&r+UA|{q5r$0(#0b|mCZ~Co|7G2QCauatv~wWdYem98
zcpv{2q7HQvSE~2ji=$S(NIu38^VFWeZ&nUo-kr;GFFZiB>1m7AuC7$2ejRZK45;VT
zYA85fhlwN;xsP75-gXO+Q#qXKO;l*&c3otfb%Ob`>9|?hRm_t<LrbECQ<HX3Yd|Uc
z;Cm2Bl33)%{^F*66=~wh9GQWm6+MhqKxyGD-1WN72A0&4-m#NqzXvFndnHYV_EH(t
zX5^CEy<6~aE}}lpXHoZ|KfP#t3k|26=;_fU*1{a<<3>VxMKqQz9SqBI6Pn*s%rC0F
z$>o3zx4U<YbWg0HGl_@!`O8<hMlV}7SmO&tbeMrh-%{veU>8v{|Ar=1{idZJDO8#i
zhbzlZQTWqM=<FTM-tC=13STZ!^ip%lwF@1k<0ko&O88RvOX670q7=Y!6!kOA$M?1j
z=4kr`_e=)BW^n=e*S^ExrbuM>Im^6eE0bhlceFg$lqyE<qB*ILY3tsdOsS=U<XsL@
z=kIBPJusY>Ke$EFF)^t7aaGi1-YN^YpN@4y_1T&}7NlW3fjQT{mpOLvp=zgXq%@(3
zO7s6BUEh9a*Y-qcM$Kg_U%aK(j3>zSxyYuPYEWL4DG&b`0$WG%oAy(ixhX!7y-v<z
z4(Iie?xjj;HPtf1In&vi`1QO>Dv#VNg7?w&Fm>#`hYMP^&?zrPdSV0)G>*mD^{3gX
z9#NPiG;xb<KQLwedz4WfDC+W#Q}zURiTyMM>6F9Cyq#S==1pyd+t+rCO&!P11r5iU
zoi>#4Yynq2a+v4Iobb?W9ezwrlDY1VkZ3+Pg`;`^1$8dR2B)F)`LQBRd-Vm^g$`Tx
z(T%P3dB>~2RPfqQ8j?qwKT(CJBHTOOL|Vo5vao^=u>IpV`6L2!-Yyg4@Ry?YOA-A%
zcB5G99zA;5i<}Z11h?Cqj%}PJ`q~^_&mLyML*;4I<i{d^8-$4d@A#GRI(XTd$rpQ=
z@#ccNh#0#O_g{oVE^r(@nEHx(o4qF(b;rJMp}4-UHy?3Lc*C~c=5o|P!TG&pvBQcb
zCHLR(L)N-zo}bP-YRh2HzSAH*p*aliTDHb+FTI${W#0P^;8z2uw~ohPY@UOz)#q7J
z$0PXe^_w?c7InJ*8L%8A*e4^0)2i1qc&5%yI`Lu!>bo8zTeM=D@D{p~+lSrPW$4`9
zjO(eKmrZ$Uh218q<=Ixk!~Xb2*1N3}d%d=E`I3a!NI6r;48t2p#l(%y>W^oUI?XgX
zX%hLGdeURdOX73a?$0<tj1$^pjQ?F$*uH`X4k4`Dq)J9LPOvPkf$G)}n)bbto%8*L
zM1!^T<nUwi{Fo=Z)h~gk2Lz-1S}^|F&e`iSZ+hjGj90!<yz!mLyUOCRI_4zgD|+GT
zy;tn_8bx|2-;cVB^-qO?isctxC*bW8U+kOZNOu<>C#%7poXH)A+qraBD|8T1y$3P-
z^Al+B?O~*Ugi&_OG$wI+$QzwkvCMu|G~=8f^^2&LHF>F7^p+~H-K(!+bln$j)TV@-
znyq|qp@OtezeY61-$zj32g#=JN<__#B|oJV?8+M@+0JR1_&L%HCJuw}DZ-hzKAcEJ
z4+Yc0cAC(SHsjdP?Np!=1@%eVw7>YRtTAA(?A^w*FxV;Tr`#^#%CrogI8(6AqU>a3
z*N65wd*kthc$Qq_EBd62M;+gfbDvw;^1(mY+7L||lbT4fz18sT_LRMR<&IMIef(!>
zAHH1rnk~ENP6wA3QN_qb5>u-)H1<Y_&@H}a5nCE5LNMibw{E4P@;k`aF6EKJx81^V
z2sC0I;NzrENWPKEMwRp;pVp!H6n>IE-q?X8gXz@N_LMCY+M4ecbBNS!F#2N(`JL0_
zzV}AZ>j&QC|1B81xQg(<38BleCD?YckqwA<qX!iR$W$4G?qfrcwqQ1zUC*G};JL6^
z0%cZ4Li;qwPr)3?pZAgs)@EV#Vh3(A#gV)w>QO<>E3TuP$%Y5IA(nO{L^dAF=i8%j
zlkkP>IT)_t0kZe^ZpfUq$I4#MSCB4RIg9RQ$6!&v(YP@2B%kKe9wFz?&|0|y<{FyC
zzF{ckudk&g55FOvj$*-^2Yk?{FiP#(6Ag7cVWQWC5)T>h{d2bPB?X&lk}c!UrFW2c
zJpitHwK(};BYK-2VuRoRjmFAVxZ1K7UQKE6oz;_^ZYA+I)_YlOkTKhtWR1nsooHK{
z3AYK>r_Gh8_>_n3rMsGyd8Tc9>5uRs{Keh3u(S@N@bzDqlJpdgA1lD>F%H;bxQusS
z7*DnuyGY;iq%5zKRMzQt1o=%p&h?I;r}lg0$^$LNQ}0(7@hoAL<oO|xt??fAS$JgT
zUvFg2e6h@B#uRKGtA?%pqp{>d0kyvRz#Uh_$fD9y+3b)+q}_={>juG&*%?iz1%u&I
z%L?SZ+JuxSx$<sGmXLSMl|;#%<O7XoqVQ4y7M&10=zUr^eJ_ojX;}ifwv}Zm_9M}>
zGlR#s?8A+|bGUN*SF$Bj79(R?A<NGDC@BlFAlByr3$45lqbJVftQ{a(Y<Y@4WUWL;
z*L<wio<{0*JD{B}Y9^E|Na6TJ<jt=_q^Qjw7?nfQ_9oD@-(lz@H0&KZq@k1hCsL7}
zW#1;9#^uaR%($is`C&cba&$L7R9(TerrVe@Wd*!+t5Nq!@G=K=<VRhP;CYrO?Q@Xe
zuhCLkS@IRL_vRw{Sq_z7nZbg8hBEta>&c-*7BqdAAnAe&x(#AT;W^k_>_r|Hdbs#S
zj2~y+l!b2m4bzE<7`j|Abyj|*mudy<W}Pd>K2V?$15OYg9i~yS73f-Si3LTMDJA4E
zOPTSU7Tfjbxn2%13opRi?RTJ?ok~CS`qA1vJ-WE$BW53eOe&dQg_l@6(tmM}yQ%Dz
z#2)NG3EMuh#XSo!p{EsPOjv}d^8@fDI|vmc+gbFqeMXkupRk)NvuS_eYN#%qAbXH#
zfhV>_qOYUzaC0nD<~`whs|26-OB9dsKa5kNf9~};#Fhuwz_?=!?NS=Zd=G^pMyzeR
ztj^?1x5!xaNPC*=yMSxxx=M7U;yppP*xlMaOtNSp{PdGp^I&CZX=oQgET2ZL-#gGs
z@w;e}&hcEio>hEoVM)@h2!0u9Zu;&Jw|+Z|ekRnhSuvf2|6MbmpP(-Ft6NSwdw;R&
zr%tl#<^hO39a^TMvkuGr{i(qulG<CFKs9b44;Q(iGk#mKcYnSldR{K|?v}!bsI0=V
zngeXp>3OL6DJNB#`hbmh=}`XbwfI~=ET+e1?!p_<4l<A1GSdlDSf0)~N|+RZGkt#Z
z(<=$t5;Z#HK9<)zRg&8-8GYW@Tq+w2DsG=morjM>yjBBl-OI$L%h7D5&ppx*J|D{O
z4oS4Mmec(%A>4FA75mnz!m@?W;R*{cDmhw(;weEm<CBPaOLK7QdpHeD9>tV1ZMnwv
zBq+K_NcYSMni;tqFNWp_JBZ;J*cinV*+}?k)?$0iTvGIzfYwJLRJvt{Y;<KTEi_1B
z-}MSnd9b#us(ckTJQn@iL!rEFnlX*BABKp42h8B_Sh{^jL$WZcm&C2A3^%4fg5R8T
z-1zHJ^t+*hYG)#ykTr;zsm^^jWm8Y9&6HfLDR?AeO#gHeX4!RQB?WD4@tsQHQ8JQ}
z(th*NXA|*d<4v5oRgTmbn`rBy=WI;h@x1=lS-5cngh%!u&9QYXF)xC?_TG)FZRxn#
zl)+<PCbFmRipVskka`A`Lt%_2KFBU0ed$hAjm#rE2V+`w<OccHC*%CGV#%Dz87wh*
z8cv_TMp~nHAa<MwISi^ms*4!}h>v*;i=<*p!2@_xgPBV+@%DEb%+lvD2aRiR4BmpQ
zN)wzMB-Uw?+@RKfItmYevnbY`jl!<CNVofSlAtq%33hGzkQl7DQDvWZ9io$+Hly><
zYkU==c-D_G2oiJJ;=QNIDCR1qUD*X*EcAvuU!&`z_559HS5f;Hjyr~mQf;SD>g0YG
zMk8k<^1vw`7aho(mruZjWP;7F4w&oH33B~6VBX3*kTbeOx57WL$h84{>!LI?308@{
zn}YPK&7QI|mi=&6vkL`|jX=Qhud>|UbI7xA0~_F4f@Na`!_4c4q-uT|g=v1|%YGig
z?5#hUL*hKiw~vZ=+OmYA-y2IDyq;3ehFBDj%V78Rcp<H*3}XhVN;Rw9FmMmYg34T$
zwXg;uYk%^>rE%1~W{1r2NF<qL6rd$!I=Mw26S~j$c&4a8mz(5Ji1$=x(+QVa8+e{q
zzHD8Z0;_wIOS;?kVfC3yw5D2vWh-38I(cX0g>^&B%9F^loW%!^O69pOFOU}%Dsxad
zh614}w6W;Vuf1;|<?ll*Vl_UnoUUErKqJt@LCk+o_)%)XUb4`Sg|4*<-W^^CjIf51
zMI8%bc05lj49f@ZMrD)=Z_c_%Dp!9akA$w+I!Siq@?ZR$M?V_7rXGs><s>%O!f1Q;
zUiK#PA-lY8Iu=Bi()5wRxRAS&c^{RRs;m#iwatknj4=`W_82?4Iv3Ll!}#E*+t{Y}
ztEk0Mm5K(~G2M_dtliWD4nyal&>>WErsg{2-z<lbh9c&ktHo36SnhUWw}rg?RiV@E
zP#$>mB<5*kKsG{-Ck<^R^&XdDJpZl4CT=wiSy@dtYKF7oZnw!WES{PwRiw>76X51<
zT>9|h2YzRaXeaq%+w=>lnB<S8-;43mzKps~Z=yM^he&(9;PrNrL;5ZitW({=t=~r?
zBQcq#JNe+EQ7tJ<7vr{>LG)1TBCi)Z=ijL(v1$KyOv+R$U+WxCJhqv|juYRDx(i{S
zbe`1Jo?+b~@ODD0mKT^zH@a%_W!Z0;wqlN0N7;h*-ottHj;qXb^9<O1&BBriS&}WP
z2hm^cFq7AKPq7_0A?<k(YS$KHbGtgc98(V6t(uUQUxT;Nc|3Qhl+|B)NV=By$SGSM
z;jtQ2|5IK1{M8qB=8gu9C{04x`(6D0G8w)V9c9n1*pN-pIiVYwEBs*ve>}h*_QME<
zMY#yf?j*^bm=9%<3(-I24*P^8iSyA@Y_X*czojf+?m6o>ZVH`KxA}7A3k84ICAyFn
zZV+pwBg$adQxTiX^C)irO@7zKf|b<OTFiErCuirAyy~wpl=g_>j;tFVmX%0CtK*UP
zBAHS~DoW2r8KAo54z85W$2U(ajQOnwyHqu)#B3}Zz3vLO5A6cS1*-Hw+7G6Ozw;68
zzFH`4IfJ-con`hNYh{m8^D+MEQ*QNDo&^dvX7!LNv{v7x1$`>nj=JYmn7fl*^-f_1
z)=$W1Vy5uV>%^YSn@$>^4`GH+h|HzEYWWW15~>>SBy);7j!#n#;YcrEL`*NF>Mx<(
zI3gTXj?NfudqFaOm%8ksK_VP_UxVRA8BMzKhTXqvOpD^rkg=CCB%Qv%z~dsc<ck>w
zgaBK_dM)?obN6*Dx0umRI&73ToTopNWS@FSFWom=<R!0QwI}LvZRJeLn^lD(^>{=_
z#UX7>S9A(rh0lvxC5hR|{HXP4R1Dc--f`S6_F7BmoIQ-ixPCHj`TG#tc#G8JJ?X)^
zUU0fsj1kBCv7+<^%y;-z9yetjjdqGf(lB#unX!jW<X3sDYZ4!}xCYfdR+Dw+dsrzy
zEL(Xp63$&7!(_H5)ue3@8iWd2<*Ned-R(gKrlhmkwc5;lhaFy~nDg5=EJ@}gIEb13
z(aroA`YctGI*kwGkJQwtU}iCI@9n~DhAQv^QKvEZS}+9^U&F*GOVBsLb6?#B{qnr%
zhOQpwue*bmy*-4#^G4iUJ(B0hKjxB&MdJ5BDzwB&lyF`!<9Y;?{pCJ{a>jq;O(lCJ
zX0PISNm3MQwoRizRaMNHIFQvI4X4!Gs-!ToH*~rs(U?3b-Z}O~-YqwLjjhG_O?#Mb
z(H6exYzouT7oL)>@hE?3!!Ml=qiG&@B>_)&()*PLrG}9+NiX#W*XvLRQ78)gmorFF
zSy|eTNAS-PH_>^uT)FJkQdln9P-bb6g%f%&$!Ft!Y8Y0Adf!Gi_Ji<zxw{ox5-i!>
zSNrMFvK?Z5;2;*e?Z#KB399-n5*}LZrN7QK$wuUU;_e&m`Ola85a}_7o?Lk));SDt
zr`tX9Dw#o@l6O+8btsH(6v#T}a5D1fLcWJ5<G8IkZf}fcxnC=BeR(4}n=N3UlK(1S
z9=R3nR*B@+P2`Wyg<{czA($Pu5i9*K<H58ylEzEoGxjKvRo7HV>a=h2@(qH!Tz0=S
z#x;%>w#85ftrhIyqE?#OL$L0?d_}GLK*;M@i+RxB)a8@`^p4!d%7xkp(g~-I5n7ZM
zd{nTwG7)30PQKH%$$Z{Zmi@~D$E)OM*1_}GlDvb~&bcM>9joY$rqCg{%CUJ<7E;*P
z4$@^?UXZ%y)3S^%)6g7w6~!0iq>}z9Yp-^KAD>==QlkfDHw0ffTV)`v*Hq$a{X0nK
z7~W(N9<jo6We1M$iGk&#OqTFy6&qH20<}G#B5V41Jn<aFM=r6zsUcxh+^C9AiQ8%M
zIA!5+w2%(j>?4E8J+MK+5&7zwl3t4rqWMlWPk!q|K~G}1MwAmd=IJ4)>r)!|bOwFg
zuFV_G?O;2m2;0V`lm;9a2J2p_6lNJ$mT44<$H`H+`BYB2N0y3q)=99~dIn#%f07i|
z7ola*6Sz%Hw@8^APO3k3DPsFqYIi}5%Ud(q{rOqw7;uGu$#_h;CN>my>nYQZcp&q5
zeV#`D8jk1**Ae~WH`7`!*fwhh;_u{Vk~qc7%qz;8#^0#qVQ+I_*Sa6qXG+n(Ab~Yq
z=}3diRnZ)H9j9&&B>RE|?Cd5L>B`10GTnu*BudM@kaM}1<zM&#n+Qc}y7rp8<R4)P
z-#SYk4M;;+QykuQ_n{RR_mjjf0i{W0T=9JgmXHFkpAd>|62{151U@<*q0t}Y%Ojl$
z%Advf@o*7z>vvM;j$5EJ_b%C)4nW|p4{XHG>9C%XgrAelU@m-vCh94c4|x*<7h5-Y
zTZ#N=#U=F8ekSUkZsV_kDfnU4g852X2<W0lYc3dKZOun)(e+2Vs~Ojk`vJ?2b-a@;
zXYP|c`Rjx3s6HgveG5ji`?uc7@^nn;Xk9dU{<0*6_h$5><{Z1<bdA<d$%gyu{>)^v
zveb9^Ar|(&jXB)#X0K)_NT+{X#PVBavmwtjF*tiaH%gm}CyJ+Kj{;wExdVnUTRezn
z&D=~y0beCUE+ru)Wfv~R48*7()ifurkRBetM~&Z{sm*8_-7hpD_j%cZU2}|{tkS{w
z&HGXOQm|%~H!|IIm#AdHFw#4#hO|LRu%E1?+@t4ymDmm)<o@jTlRL=E$tf$#{Xg3P
zxqfnfz6-op`*^N#_nq(I>+8PM-*bWQd>?<mmHvJs#e=Ioe7tR}wB%g==gnC4&k<`<
zm(FZ2j__yypN}|7PW#^;am3;UehWsfcK7vI;OVh`fuA_~e2>L8R=xjC1B^zEF&k}W
zWIEQw)YQntc+_aqQAUP_h9+Z;M~@yo%5b#tC}ZQXV@Hks(@Lqe&EI;Wg@1PY$?g6$
z_P=^&+^XP?+2a16ec+!b{%QMn{q(2NBUY~P{MU#l-`(HRPDJ^$5C5lVW5h%MF5-wK
zKE7TH{O14X8Ti-0q3mfdMG^4NKIoqU*ZePm7q0Vj_wilq?)$GH|N9lA`%fXAL@)ij
zw*KRk8UAa$|9(9T_@{VoiXHxKy#L{U{_m%5)ISBh@xKK8|6f%0lNA;9|1*wu2ZR5L
y_@__*Z2$N6_n)&TwzWzf{&6ebr}XEK0G0o^|L1A@$qN5?xuSU2pV$A#YyS^`=ER5q

literal 0
HcmV?d00001

diff --git a/test/models/data/transolver_irregular_te_output.pth b/test/models/data/transolver_irregular_te_output.pth
new file mode 100644
index 0000000000000000000000000000000000000000..ce9c394c0b592eb5763e0cb529e196799fe12fd1
GIT binary patch
literal 199147
zcmbSydo)$i*Y<Hq$UTIR+;54J<m|afl4}XwL{t<~A)!L<A(e7Tr9!AANs*jASCUE;
zl1e3X*G(#+Qhn$5y?=h=9q&87J;oaQj5GE*Ywo$$nsct_d6t{AkT8cMD$4nPE<FyH
z6A`|4%eJi{kpbbpLE+&68@Gq74fl-*@ZGvSB6NGiO#ihJYfVBoh1lE9<#_%7UGajk
zVH{`9`iQOJ>jO=Ew?_npY!mkl4_LQ7D8!%N6~Fs%-^f|wVFLR4D<|@e+{S-$?Z$vG
z!T<V+xP8dhwGkfw^`&^2&`JURGp1p}D~0&?^`YCtCb+(ChzJu|>dbK#cAaPHEU+hR
z;{Uz>=aXpq`t@5wLIT!D1Z~~2O?*kXe?WMEzg^J!2=Op61NSiTh_FfSVG`R++?*$h
z&JT;1IL-gp|1`J%)@>#n_@CTnvSDlZ=Cu*Ne7A1n_iJxE@Bi0(US4MZvFys1o}~Z0
z_kY@dZ<#*F;D54g-HwQWt^aX?y{*pw&Zdx{h=`DY{}_=1uargj2K=Y}_eSV&^!_I!
z{xhp)n%diH{ckJ~NM4Bp8!4k=Yz7P2ROtEphwVE-F(g?JVh=oks))UKr!)a?T%JG%
z-cG{c83pJopbq*gw8)uxa@4@<0_IsdlDcWj$*)bmcxR3xjUE9qYjP3V+?MD1L`uSP
zgGnT#cnucLm7qym6o_gI3!G|Cb_IA*Hv?%j3_Oi8dw;_CyUq0F-a|}uw?0HT%F|)U
zCIbVX7_(`5=&Ae^8OP&9BhiHsPEJF+z|ExT#5n7}(t~DOeTU$kvuOIp0?;^o2dX#(
zU;I)crfnfuudT!g9Zv(#e_!G0XHiN`G@v_47EkVK#Dlr*7^>reg8k>g=|mTP-Ij%!
z4dT=?P@j1<Rg{V-nq!303D(n*PzS|QT((geG)0;*w?UCCT&PIci{r@t+|KwV%R{_L
z1~eA*V`j<&XvtoO(Q~|sy`w03d@BJbhwkS(y*!KhR|V+J&@y;b{{t=V+(Sj*{p@(v
zX*?jCfPdaDqEkKGC>55V)AoF3F6`u?{+dkuee?@^P$mgFb(WInpW9*K#}JfuXocey
z%kh!nAxP9J!pp)Z;pxJo_}9n=LTV=>t9Sy4^iI5m&UmOk4YtUwA>3zL)WKsR=$kGl
z`|Kvsr~}sIsPHW6&-;!>3(T;#X(=A5ZNz=wj<a)l`J~raA1mTCm}n1mmNvbF>5lcN
zu*Q&@v~D79+g+i0@F{){Kv);3KnLV+VEJKBB3Y#ZJMIY)yG6~c$ORYFKXefuY2AhL
zE8!4Pq)PP^!$?$w1x;BkLE=@SNagLN*xPE)92Qa`b0bEuMB+AxTjt@-7A=Ta7m5p)
zn3CWgTNHMSVFR`$g5S(d<l2VWR3Su!{3-c~+H!+za%vHCq1PJJpGdHpOO4PW;3EjX
z8A99^ix<X(VA=U~cqXxh=iO~d`nNJXkK2OiX+wFLP6;GFC>;J(=(6)=-$0$uMizoJ
zsNNe(Vrs2S1r#grX0RkZc7GZ*XVyT=rzMyzyaLTXUdEN-3wZalci_d#DMX=NnKnGG
z;yv5g2^&~vGSg8C%r|votBoAdaCU(+RfTvq=QrCAh1_?U1}Jk&l~tT4L_z`*ndi^r
znEr;}JgNE=(%k)xHx|Aeu1a!f{JH6jc<d20Nf^M$)+P)niiGy#ee49U)2#V*dlCs<
zsPy70J1kVm*fpeJ=RZMOXhp3uHrqq;%==(?&!3cU1ga2c1l_Aog6!4BRHwe4o!PP%
z3ynM(*ED_l>BM3h!I{cv&)*NChaA|n{&IXfG#8@I<P#IuKPa<!gjIey8{2eRAX-SC
zHJv<@*|S)aCe<uJrGo`rne{C9KyV~X(%g%`me`|@*i_1yk;%Jzy^Gl=tw8n&FCgnQ
z^vJBygXD9{LL&8K5D%`2VaE*O;hy_1aQ)2S?jMurVlx>GZEIzoq<@5)eG)KYUjmtv
zwxfc~akigUqKaS)b__=G&Z?%fc4jiHd*wwa`_d1`%&(x=^Fcf=9g2?y{i)la24p&^
z;!K+&JSOjtUL!noQmbP$q<yUhEPZIJ*G+Vvdj{3#i_qa<HCX*Rn-oMVQA^oEynf>V
zvv0VCFq*R%L%#>`zBi1hnq`t>)*>J}qXw&%M&s=76KJTH5IL_ug=$GvGMv-CR<4q=
zw65?E@6tjMs-<)Xs|N0J<8hzW<{~NFB&9(tcK<|shYeJ9qA_*5J%#;ty9w6lJYe&W
zC}6f`J+?$lp+AT7QE^xvzC=ucV@Au-|Bwc?yS#?kWFbI1X4;UrW#&-taTi?HPQuG0
z3#govA?aPWlGw~T2udZ($o?CS#MNpeVV74k4)QZe+|)GIx3v%&UP}|#rwkY`(t<V5
zOo-NH7E))vf)6jHXy<`O#%-<&S7amxjcsy??u0CK%{&Y(w{kJw(Y4;j{|#8%oJIR*
z!|4CV3V5&8=>DXQq`~7W1m;PSt3Fv!Q&WtpJ&mxmIv*q39LaT_IQ_OF5q0(((}nvy
zu_0p@`)PYE3a(p4^cN(MAKD)D+b;>sTKf#k#4e-rr5@(j-cfufkdCKT+{LPmJZ@Fj
zC6GFhz{Dp#LPbe^vU}G_?D8?AQe9o>A<RJZ8eM9eejXI3_h7Bi9EflK!W}7gWY)CB
zqxe4~+Mp~<XW3=)V@(Q36xM_0!Smp7dKB*&xxwMCB)n2QgH}CaD6MXSJ=;oQYoIBN
zw}`Q4rNrpecNI*h!bxUFUO6^=*k3=e{|{!zwJ>L*B+1d8wOFe;3p+N9V%@C!*z(kb
zoYm6612P7Tlz|XzS#}KLnueLcIz8MUpiKQ|monPYhw<&QmrQ47Hxs$t4)?n}Mc>XY
zXtG&ApXOc!x2{AKgo#W}oI5s2X`;hZ79H5FOz4tb__r#ITnv>aRv%>OujTP5;J1pj
zw&bGBT@hlgs6<PqY4Fp82JPTHhDGDD&^;vs8_qe9w$VW1Vx)_1+4q^sm`HYcktK-}
zI0elG-k?5pkO@g>L-~7au=!su6UZNXufX3B91Or|>bAOGF`4YO)F$JrThYT?o;2sb
zLd#(ZBDFG(sQmQ6_pTcBgKY|`ER4c8Cc^OOpA2o883DZptBG}$0lQGZ0_#VT=*q|+
z=pC*|Pqckt)@d9^Q+FR;WKscCyw66XNqM}Ms&2IQo<OeOx`^$iG4RS<m5$elv1e|{
z(y}f0QNuh5WA$o4W=b90?B77@lRYqW(Ruh?D1~D`@|asktf;Zq5lpH2g5#dAaPFuR
z>AjswLVp=y;HyCLvYmtR?Z>PXD^EatYA09Y$7c4(_F`0zRv>lVNpPiJhYVeIht4El
zsQvJpIl=wP=mw{w!_IPUlA{0(jFG0}#0nD6o06s1wvrv=BD8093+}5{q|=U>f#2t&
zM76MziJxv@<(nMDj_Djmje^UlvRsR3yr^fbTQf-FFKaw7YyhboE(WjlLET&1nUI+#
zRL}MvI+@JHU1Cw7F8c&u2wVcs?n&5vQ-SH2wh(H)LSZb|ffcb3qie6Gk}94G%?n=3
zu9Uxu=k};z<%D-U_EHo43(<#?J&rIe;Y@<mR^#*YGDOhqJJvaF#fpDoq~wAeEsIhi
zS(SUB@W=~Z^44o?=Bqcjd-4?A`T9N->+1rQ*@>|ohlqDbI%wqhz}=Zvbl<TP@SvW;
z^s}$vhk_p+y{bk+zdnJXs82BApc0-C8^C-Mg!0~IVrMM|-G4>c%8En~bC;tRy17`j
zj)L>sTIdVBj~pF&cI?UsEI%tk&ZNGC6RtL7>V!+!=x)m_pMQ%rsMREa24|R+T2o2G
z_EY3+aTB_<CJ>WNrMy>L_L8-2S=bd-hu>O{qlok|(pI01+lzl9u^hyK{;yzAWk(&P
zjG*+gE*+TK2u>hHEK3NP@-+`R4+OA&RV4~Px{n`BtSD>O2jd!lu-QY3USy82t|xC}
zt86CD$ovEP4F*JYy&15%BkYtflkn}xP1K15Ea2zV{?uFbGQ&%O-MJqhGV-+j_e4?>
z8;{9Fz{+-x<HwH=P^IG>+hn~N|9X9f?Jj3np<`(fO8U7L{vR>_(n?y`H<8NOM3Xfm
zf1x4xLVb6=Fn=r^iM)C%@O=H~$7ffdOzi^pn9L>5BI+<NK^8LJ`9R&TEoi*!44zqX
z8~RHg<F!>vEcd1{X^>mOe4i9fhWda;uI*tzrMR;F>xIatSTmY=Mjd0uDzK{jC@b?%
z7&nF%A{)>FpMF`9{_Ag{`{W2-iHXLX07>l0kRV!)D{$V?H0Un;g!ElD`P(A|B#h$t
zvP6{a@Wjg@Ur@B-I``585o$X_l<Lo%i54$}=-DTx@O`^7PWkMHWkUMIg}*L_JVb~*
z$AZY+NdVjLqp&Ph3T(c=#&tv)o8L)t@5a7Ihs*nj%akE7D!2|?Cd$*Y+wu5w>NJw5
z-3gHgS20jRX!LU)8k|~(zA{1Z;pYKd|8@pcIZR-ywBk^)zzgQBQ=&bKWU2AX&nTla
zjG0DW6f`Rs&W{x|S$>S&aIu25tL<f$p3O(_OvX&{VlZuW<R*$Or;%zka9t`86)*q6
z9{<ZEFzgHluQS1s=Mr@B##8Xp@&YSmm<xuFHj$kr_mSJxfZkc6bc@|=`b?w-2d5k*
z!%ZJBE8!eh_w`jK;+GE=buNdYf(G84xFf_ZKOJX3T|j%hmGJf%6-woMc+^85=f5*2
zH4TZN>ZD9}SGS`DtAO$G6;S*pll8zLva)s%_lFdr>6`ZY9h$|k^Sd)Se_D^~e^`%d
zC%ANB-wxvK+stMN-ADW4v-nZn2hN83)1asUv|QD}p3=LB4_oxH;72}2EGU6HlREIh
zms;FiRg3eUBoVt~`ZPac0Xyq+CV0-;!u(DC1#)}WQxUC5oHE{ndS}&W;PnvfJM$df
z4;;WO4P9Cgb{iWj7|5LCO3H-R;Hr`Zv|B2lyFK9+JhnNE{zBUHSgr&0&zsD6%Lp)U
zUQMR9<I8Z?d=BYL)1>a_H{+k8V|b&h6yv$Qn3nwm-Ve)AgP{X>^RY69{n<@4XDZ;K
zDZep?oq?6ADbRkY3cnq-paT!{d2iAv94`Bdr9UT7zuSB9=O<l!dCix|nX8cGIm-Cz
z++><n+svq_oyG36x7mYXLnxqp1=Z6esLMPP+8=ou2Qs#im`4@xplb&@@b1ER7lV%j
zO=!kxE&AF&9JCwH1E+_r|7ev){A_)Aj-L(ag>Pz*9zG2h1wUi$emRhh_om?J+of>k
z!f*DZwlFPy{tV|D?!vZEN%|-(3GXQ<u!UCysGi(eEN@T8HRr-$$=-FixOfsS_j5%3
zo<LZq`~n)TQQq}98@ld`97d;2MS&kYOuBOrU1A6E=C24CcoWFBEs2DYb3m34q!Ck(
ziR^~f?V!4{o#}r}u>Qmrd~s2StQgdS#y=Y5R&F2&I=Erxn>;Yhnh0-PrlQz0Eo>1i
zV5ezS;o@1yXmlha=UhnrU|u=cK3Y$nzO<zZ@rTH-{;xdGDXV~<N+;?uhk(f25r+jj
zq+^>7iMp#x#7^GBMT-hiXZjPo{QVBxue%5Pi_PeVzbEj@H5p>s(87HoGJ!UH(q>KD
zZ!(Ktgp%cZ1)yzdE9-j77y}=AlD$ne7`|8+qg|A7=dV9-H<z#?p*D>7U>Bq6`wBFp
zM<He;31TpT=$w9t-i15C;m2df^iU7ml&{M;J*&d}j>BZ`gYQ_~bso32h43nd73tX@
zSBby48Ld~IM<jLgu=Tteohde*2KcRk@gN;6pQ=Igms$|z!DBdEJ_7}9tnlse!=!-s
z3*!%l*FSU)CNj(*BO5M7d;iMO*Vgjz@e)vtRpJ<aSA&|{mY`PEByxA!8hliwj&`HV
zdE>VHT(dBqJrp8A4_o%&BXdEh`?HX$x&hrgb_zo_&BBr2`uIe9BGon!A~FRw<lBo=
zjH9I(5wo%)2M!98BQn!ymEl8hJGK@#rR72B?R;(zlZ0;*E@1VaZ*Vs-oyaauBBq()
zs5kr&2cGO8GFL<y`Tl?G*rD09#AgW8+>hf(!Ezu~%ki3?D_wXd5*HLm<LrJtl2>Vt
zy*F|}&_Ewgrgs9zRgQ+LH}J5p4{dV%xtU=-D6oDbb_+|9sAyHxF%h9AzM@2~QG?`&
zOrl51KSQREC)8bFaQzx>x}P{x<2|(~^VJ-m+_*_z{x+dGj!#g(_b~Z9aEvv!zJT$4
zuj>__6Eyusm`F`m>RqIY-o6v4x|=((Di<c^_s&B55iQjH*2)@}4X|ylhIFOXBC_OD
z8^(p$QI8=_QnK<iD(qN5tHoy0f^+@gKUIXt+<1wFTb4kq@BnVPuM9tn8_{9*arQf}
z9v{!-plXRJGc8D!zTKftwMQ+f?*3#jUo1v#mUl4K<13l|0)7IY8$xIA)F)T&T*W;v
zgz4e#4%Va7kWM~(j%8nqP)j`lY_iSc>HYc)UWGxpW5!i5*pf^dZX`1Ic!pFa!H^eY
zbqP(ie?TU`_HNn2uRFFKA$-u56vh3(=A*4#sn^1AB_k0FmF02ctSHF6evC|P`2wkF
z6X>!&51dh2%2ch;q;~l&O#T~H8WXGz>r@vJspclEn<h`{TvB*jJ#ule!+O%|JBvjA
z%R|M7<BZO@16>~P0AI%>X{qNiwDJ3gP0B{pz-j{t*At{ASN1Tg(!R3dVtYWz$_$wR
zQ93RxLiM(ZQ)b=;c7H}B6nC72<`gNa-&PBXt?pR)A_PXBJYqyMx*#$3Fxu3Gvr|t#
z$8xm`tcd<R;#N}3o=g#=i`i!&rF{TNaxHj<SWv;d^JxC69DJfDkZ_X+kQ*gLEsI}s
z&2C=9G?|@H`0+U7b3F|~R1j5iO`!O$JsC+m0(&QVgHXE}e%~U4bI&9qF@Fk*qpEbs
z{SRDOWk~~lp0FW<fz&Ozf_tRu8GaCcjeqNvsZ(1aUJV$8Pe%h`=-zBzv%!7%lOhZw
zRTIdNT0hS}V=pf$?k@PgYK09WsZgjLOuY~3&`Q7Am?82K3Rb((mu<RqaYZuG{xFdU
zlpF`o_?wtm%m;r0L~y!f5J*0{iy_Zk$&&j^*#!n0DK`CP7jtFEGs9cBMU}&{KNZNe
z^=70(B99qx;IQ46*_f{`4@%5560zhO7(bPSL1{axrE(s%XWYRHqz;#}BN*zc49DkR
zXZ3GRq?|7iOxpw<8l9R<vQBA}Z5IWpW=R9udwzj{qyM0^^9028|HP(nLW`1LK+UpC
z=;!Z96m7FnCCP;Te3y<tLoCtVs+#rd$>M5UIf6FIZ5Yn|j9*l&Y3=z`2=GlMN9A*%
z@7VyJQ{nHi{cADUp&jkZI9Ow#Kx2!Kkl+y^(y=;+Q8}`h9=<IGIU8fxr&c$a=JtuC
zYsD}YFEB!lT_$w4@l2u|6a+Fc+gYpB0dDVoF{(N+AAE#Mm=3mr{S|cq>)az)>&sa%
z&bHxjqdRpNzsh9#Yk_z9G$Q)^9_uR8&c4ZQKno9VHs4?ow`FZA$_TuGX6<C6ILj28
zO^%a+p6R5bwGv~F27$%*B<$OM7j0VHc-a*VOyjldc*tQQz727tg@ZTi^UW7Awmt_j
z>E{%>`(G(=x_NcUGmA*S<O3L5znNG3u@%mqx`=vc4PHYBafQ_~y3>6(#K-v7v$CrA
zYpOPVCUp%AH|S9KT7mv=KB1oWN!-OY;f9}mY=)dNWM-XaeHN?GbMtc9JF+q~IpjDa
z_p}FJYUYtp;({8tVtLt7MdXH=H<{huf!?wlFxG8{JW~;>dv+cvbG9I9apM@Mk&e&i
zTH%%uV{+tvBP@2COfMM?U}(%%a^J{^Mp`Z-VfX9s&y2sIx}t#1H#&+N`kWxF_&pTT
z*IdP0K4e%gi_9J{Au{oQtxk`XVCxAr95_9lp7}SO={v?H51*x=#p-fgxmlWFUHNsS
z#0XlPYht&T=+g007Js=t#Y5{On4)W8L`q@?9=NMSTQ@I(>spT>Enkcn@5lhtC?V#|
z1{;ze9FL6$)u=}C9#q{@iLqKN-i{T-IJp<hhF9sljIu8L<adVYzdMhN72QVjqP@`A
zI*+=^e#83X69^|O%*tC@04!Y2lIjPOu};m8l`XEp)0O-hGO`2PQu=UyY&0=@*o`p*
zd$3F;6jj{C=?h3iZ>=0QzU3V^Bw!|;qajY~u3F%|{zPUrX9PE`v89*pCX%fZ=FAH1
zEPOQZis7iLaJwzmz)Hg|oHXbLJs;c9Kedzz=9pmiKX>eypvHK`zd(<7aU@qln#x>T
zO;#GIlX3m`7=G*&iiKywEY~W~HCjvTAG~4nWUBD@j;SEstW6HzJB<?}_QCSpPA2DJ
zDoVWzgQD+V<nZUuK<5h4o+Jg-elr=0zr>;a>Zi=+zkR5pt3lTP6UC<CEnt7ujtz3l
zWv&JfW5&$@2+a>B&dJZQAz~r@^lps#=#xqITSpQYTLen8)#>iGPw0~70#2H?aJVIx
zAH%odZ&Obix26Pl_37b3*BRt+)JoDMl|s!m^+00e4vOTLL9Ik3{E6C!je&l^e!9hk
z%@V>buz)%(GldWF%9N{}j9I%+*Q<w)q5j4SWD>e~qFZLrf~vh(TOkQ0{Ft5|{SAY-
z+OTbBJYkOC0?zE?Rx9mKb2&3AfD~%dDq6{VwDTKqh(MjR-v_Ww5g_~hzBBr~YY?1&
z1-H~nP*a(S>?aVWlApg}k5n;Eep1EkEAS!lUpm?81J}@8QUUhv^CnyDZo&1uoxom5
zWq0yaak)3pv3YA5MxM`WOtvM@f7LRpEEi)`WF<4{^8|8d^d?+0{DQ%*HptW^z^%eE
zrdnV-l*X2W_0eT$7CV_Z#xKUm(oCE_Uzqlon=!*%+HrrH7B;+|NS96%#YtMf*}|++
zo_FaB&`1j<g|`pz&g&h)!Vz5(+j<1O7pc=;#p}4k`weu;FMxtQYGnM_LAEl`fa<T;
z0H-%kc_IV;r0~iFKDLlbtb`5`)x#61<-s{vHm3+QZYi?++$z}bFDyYU@DA#SW^>cV
zJ3!+{G~{cV(((Nh=_A26xHxz=c-%Y%%=v7#)OQY*UosC~yp$sHm-E<#zvXDT`%PB$
zf-3D*)1(^aIlR~iEpnc#Nj)sp@R8dLI-KoER1UgP>Fi<1>l?yL<~MP`>^EA!b*3u~
z&w-)&UGAAMZL(v}eT?_L#GB<H1*@<BLFvS~<WYAFM8@yr&-a~VLW%?txY>z!1COG1
z*exdF?`pDO5*O50FC&`=F5=^1St_*RC3Ywk5&s!7bXJo98S|P?(q2;3>Dr53>tC_d
zaWNk5%SZi1lSqR$iz=(860?lepqgz-UGB7_WtS>TZ8M?k`vIaV5kzL1o71l+6EVrC
z3)V>=U=sR@ahkk7hQ9oX#z{BPZ>c&r@5u;*r&i<YDW4fmVNsnoKQ88p+`@>yN;WwC
zFn)oP7+i6hjUKk9juQO+-`zf*VqPD*NG-+i93yHIB*jg3IEqp~)0ta#O%VRx2B%LR
z#^~prtemA7wT+U+6Ztpb!C4FFedYinn}4A|uOJFEYm$p`wV0>Z!+Wp25;J%+=!&W#
zc)RQ;_Ro9-&I`YSe`OsE%x!}S>oRa^=O^s;FULDWvx%xmCkAR6QszW9TTFmaE~^G+
zi5M{$lP6n+XHfTRJSe<T3upMbqgGv=`Xt$qn6W+JLPMF}I~=;{#BETjmH?%eQQix?
z1vF#MTa;EgfcgGad~8{aI$fSc{q*$l?q?~y@BAHXs^jp%Ela!^e~^jJ|G{udk6(S4
zDu)_cVxXz>4<9HeqbV7Hm{Xegq4Os08C^wp?B>w3%_s47{!u=jqC)MT{li|tG7MU~
zo;V&Gz``s=<a|<~twKvdL|P3WkBgG1Mjg^`;Y|Y%Kj8CQbEw;b7H&`bB-%N29@pgb
zpvKl`7@GQvS6p(E%zgF-O7d=lhfM|~T1mobOHcCP<rUQGl14ov3SCPUu;VWbiCM4=
z-RZBwY#l5_XTkRjr@ncf=F$~7^xzoi%RXnlbjHAs7tifDZidCZ_9*i03`VSPVO~#C
zMkS>M^qf>Y{5{vm-rD$%-6!w`|60waISEr?eAgMa(UU_yo!`OUjeCpjRF-}!mO~5O
zN9=&yEEqbM%v-O%1fAmqNq==AcpKT$4kJb0PvO(3r2HFZi7p}Emk!_(d3joKT8erL
zjk10!1mn{g>b`22`S7I_2AYj=*4}NXLxq47FGT{&-}1I@&&M|FTx{M>7<x{U1_<^s
zq0RH5W8M^I%ikd!nm2@l*1j0u;DVOBY#HmnWiYKrjD9b_f~ybCVQ%Xx6Y)=8RJcPD
zHDZRXY=1~nw<$%i;Y$?yojwXy+E>tf(lcCfp25m(b*%KxP4K}(9+huv6Pqvw)_t7M
zifuZ8<^@1>n|6^c>x!}O(^g1T-;YK&oXLag8oZqM0(VZHNoV;4ld0YZd3|THNz9g7
z5Kr#M@WR(@u=5kP;dVBb_kY2rHAP4o#xZo#D6I2Z51C(6Ny$-XOuO+1hMz>@*{{=x
z>-kkA>+?q3?Ica_4&|f9g=LWT^cr_0d?n5<tU;Uawmi}K1MH4`QS!EU4XyvO5VRaE
zU}IkvGyL`hANS71zaE=OK};Fm?pTaXQ<aJ6H9d0WOC1~I^NsPo-T?1ZHqZ#K_qg!E
zQpoo?k6Tijq4GjKIz4>HnmTP~weLiso7QY9DlQ4i^3}+CPNHcC8`%z>*T83;@T0R6
z*(r1aA85%Fr<!vtKAK4*zbZh^ri*Af-oYAh7vV~+QSPE&t7*Z@^9+4=2v?SU#nr_n
zsD45nt#d7?`?nIjZX!>;@2b+yh=XM5x5Z%Lkj55Xu;R9rg<@B#J06>pj!L__!7ECi
zN(#Qil?oC}Q^!XzixFm2ZC$WKPzLrytI{Lgiqu0m1=sYdQtO2;(Qd*Ytd<{TE93HE
zsnT9_xwwxUeyd9gN`HY$`3WMeIG?`U;6m%ltV#6VFZl4C9NOJCph}g4{9Jnr%lDqd
zwVo=Nf8{*7R~4~&Uxyg`$Qr2pb^v$sV@2raT`0dHkr=8rqm%7tEK}6R1*ezL%YOw(
zR*wvi`y><A&8UQCBOyB1WewcXE9S4$N9cFzEHC<z5;}ZM#66`>#7Iq=Dzrvpm54Z1
z5-NqOgMV;UqZL_6a@n?06}<K7Am&%jplydA<K4I(Cf~f3`=Yp>-5LE4&&PgeB~r}L
zc(Vvkwd*Y7fZ60xLOWW$)~9mQdKs097mN_U4xW8RhK_Y9u);0EC^@Q0Uo+>iZJ98w
zDpf<vqsHuXpZV;NsXp7#QCi>CC=0Lhm1!_vOCTyb&dM+M#joxeP-M3pQcWz#4)r8N
zDP<C6p+nT|+$pCk1t+Z4rD_KRh=p1(v3U20wQmyw+bL_w#UuxMN45?$TdL71ED4UO
zZ^b)#!|dyt6R^cz0X>r>$e3~yTE)I#+eBp8-kD3Fd`<un7BMDU);EF;r}=8mv7acX
z*o@x7?^y}wXwoy^7Trb~!D`(cGN5ZkL=q|(3Gb<x_U0#dzw;EDHmc1UUP@r~`It?e
zZUx$zg%NW;xAXhne$u&N5AJ-?gO@$H#NweKy@$4_?P~zd*Axk-?g2X7D@9bwfB}tf
zjOtsy?;Ub6=WZ1fZC%T@ZFr0p0ez^kNR}M=5eH8lKfqYQQ#fGR$bPgxjH18X$+cB1
zyiVfJha?AP{cJnhZP~=s?bjfFHvK$jDM6B&s>|o?eCZ_z859~d1NkQ$nn<FEV$UhU
z*U{j$#S&Dhe;Un7Pr@~)PhtB4C3O4JY1J6A7ZO%mle#yx^`_@saYkSgeCyqaeaot_
zM`90a7_o-5eTX6&4>Ea=c8r3!Uojk%1fp0YMceMY#rgZIAuIVTTv!o7EPS>>>zWbv
zlk_L#2n=Fd?+^H%FozUARi#e;&1ki(ok`SL4%Fo~EaxSIaN=(uU#;<`b}h!=FhHqo
z0mLJ=n6D?$#VKdc;GlLdrUwhrIyEUOu-KAX&fNrpy5(3bc?!j4m&37zD@gdKpIqhk
zwPerGUNXZj1TC}-s0{ZR%vPI9&fm{r9Fw=<l-X^#*jEUbUa!H_jP1OX?>aP1^(33~
zlLw8zil9d7Ei;L22bJUhF!Qr1jUF83at(UXyfK{E{9O)~nV*0&wan_sG99|KG8vi#
zcEXhMUMO(vV_m$q!=Z+sXqTf%rp&($M`kLMfIV?!<mVjln3ad$3a+qXN+YQ2wVkYc
z=Ybv4d0cn?JIXm(VO&=uJL@q&x5#*qbvj(So8&_N&l7CLkw>U?Y8<-PwsEr_En+6J
zf530k8f3DaF#i4k8)?GVg@~SmxX?DNkDEkB7bJr6^HwO^mISARm8r?dMvTc1X5&2O
z6LMGvJRi$2@&%VrWs^L7dm4bDo*&rZmV4-yXwF=?!=+iHX6)ao>BxEYo7u|80PJRD
zvevGaq<SO~ZaZ3%+)JLcW91+u=WHh5MhnnEegV;suVW^}SkTSO;vqFfm#j;_%*-o(
zf{G>DsF2zVuD>jq-wtaCGvOkx$sS~vrTW6B3u(Npr4b}Hz8$y5T*M}wZKNb>4%JFF
zqAzq+(7)pv?(q^uT|SSfVfDBE`BEjCDxQFrk9E*u#St935)2ttw_$XyB(1E;fGtLe
zpcXQj+U$<ywryGt67u`l@bCFBy4aTnpBEz+R2`sD-;?*m>J+@27D%*$W>c4_Ve~AW
zLsLf&GMiiGvv7p43LckH(a#!>J(~}z+x9R5O5ZUeT#1JC9EPf&Iy}uGRmgbg2=m50
zXhDJtWBFK(3UW{2^hY76C;9{@8tp@~z?rDKDGt6)7NP1{j?k@h2emz{>3C2EGRdB7
z$TbfjMHK31ZYJAur&H5|3y^b6o90$%pm~oPnX>&i3|yH;_N4>>zWu;VJF@~N@*=_T
z<W9!=O$4mpBTplRX9N473vU#N($Ssb)PfJs4Xl-->2}I=d-OHzzujbYg>j{qzXOf=
znh5UQ>xoHGBWSD$A>H#g@d9sDq3Tvc`uS`m`Ken4V*NH4-y6qrT#e^t{8fg?c}mn?
z=N!hwMPZm?JS^<sq0a$lWCJtt_9I!kZ-o_fCkmqC>h+j^Ru9LU4>MkevOvGD2qQk~
zktv6sf*V+}GVR)!)U*N$C$3~&fpRzOU4SwfDmZ!PGu*j{hjp?I$nF_~e;4%Vl@J;F
z*{h6Ql*FMYQ-z65&uti-egU)kgn7Eff;2g30Xi)mfSQ3I?D?re7p#2;fj?83hHyQ?
zD*G@E-)a3|kp!7fPeEbyDqff7735S&vmK#l*|e{&WM#^9raoSn6y6hIhpwDN<7qFk
z#APpeq@_mYt>23}KYnAHo*HI1oMx0-m6-nUY#6GMWdjyQK%DFb)R)zwH&u2Lg}SLU
z4F$;o!&Ok9`4u9r_V6_(Eih3i3qSj(fs0lx_N8f%QF{TfdFsHs^|uplPJhWf>F}X3
zyZ7SAsn^(SnhrVqdE|8<7VS0)(y=8BTJwF}yek3i)(T;cpEh*GX;Pbz4a7=nB8>YU
zCGWn^AVGmOY|C$5(%-ifKinB$b{<W{c)jEGbG~Mwhj=>a{(Ke8Q=GwSe>$0{yqwgW
z-T-5(i+S|ML~8p_3KM0j*#74uV4TR;Qyk@M@}e_f@h=W-dASyJEhK?DRp8^+W~hZj
zVCrhbq#8L<85c`-+E6*<=*ZK7@|DQdN+4w=UR3D!5$5fWX|%fd71Q?M8^ZhdaJBS4
zb{_c*@!h+vWY4st>Y_-7qj;G6C-E!P2>ijafa74at{yo_r=Xg{g=^0h(fyn|=KXHP
z26ua_zqcmPmktT|bbA8)>q;RD!{$->ha<?b>A)FI-|^RlA#^u2BAjc(n0UYrZ@cnP
zR=x~6caPz3FMV=BOM~8BB)|+Fb)w@x7EoL38d&bL17{>9VCuLPW87&?UP`k#6fwv%
z=Y~Os&q@rkP$YuaPvGW5Y2dv~m%iO|5>>9Q!Z<ZEG>^T&bWX}5i82kSTF(%#<<D?b
z#~EsLCxP?CcQ~WWl$2GyVDkKnN&kdb+`o~tNObpYuBTuz@5b}p=>6b6)|VQ<U!RYx
z)#-7Jh)Ktg+V||xaW^(({%;h?j)Q`(2@u(O2SYF1NA30uc0_(AeW||z`?_VQ>AUOP
zQr$VU;+PZ_-d)Iq{<dQ_1#AL0-+H94%<0ef?{LFfN8H}7f%<J8Bz4n5TDYJT<#)xA
z*}+1jq&ty$r>=rksj^IYhXy?qKAGvsoX;4vwZUFK-Xb%17Jis7hC50tkQ=-QIcZ*X
z3!l7T-@ILd9)EMNEm@E5S2+cvr@0_E|0s-0=3}3aF@39Zk!ckZrFF-@q9i?lt1*qg
zw;09*=kqWtW+v;Sl))ZL?_s>-pJK;K4Qegl#f&`Kf}CMv>^Pl?k{@kR+A@~h7}5Zn
zb3w#!TP{|=^Cs(V>p{cMc5Ig8&_nN<a10l-o!uqyI935I3v;p4e-cWZbm3WLpFoc}
zdh~Fc0d4p-pZ4Dv<!NqFMCW@OaO86{%2zqlu^p!&`d>fqqjM>rdzegB>K9`D<#pto
z?lIgcS535BmqB#xYPSB%AqegJ3Oe(TGBw&)@eiksSt;I*UvlDc;Smn`z;U2b6|>p;
z;K#7*?GyNZTb=Y=nn$YX4T#JyW=%);;Dh-p#MH3}Y%Yt|n+R_}Mb#|y|EEFAwLCCw
zjth?M=WDDRRzvBU4Ct4wz&$?>LQCB^j+7I6hTTR!DH8I=O$*fxOlXjoE;Pj%QUSve
zc<tYU^@UDk{P1MzW)ua*3*_l{FJqb!^8_9%I^xTZ=P>Kxc6jEKK>YZ1)xG=oFsxCO
z+MRpN&hO75(E|#oeB~&77*(XFj@MzA^;u9fTL?Cd0^~>9aeT4%0=6`mgVt#yQd@8Y
z0(YB`aM_dK`EnH%^?nUjr{@u4oiRqvqYeie-$MEP0&M%D$(SauVFqd{;LrSA#J<bO
zNHqg<TOHH-qTyR>I&b!IRchbmjm()UyyZE7+!-2llc6+TPVUAnFIF)dV{6C*pQW_D
zFNxRRF@f^bg&@JA23@j#vSZSJF-QA491?uRSRdYvf0dG0&!_Edh=MW&F%H$N{e`Kq
zeuVe18K%#XB2`~9iQo6<DD-nXI!f4KUiuGQ@boMC-*;wLXp0k~eT7Xhe$43i`J$gh
z4!#!UG9#&~wEx#;-o(Hdh*x|D69WhFO~3?rqpc5CN{8X`yN8(i`T*lqG8g+_36MJr
z<UqMznr@sgO2Z#@W6e!HjH+)0wbS`{Mns)9T_0j%b_>(*Kl?$itQP&Aq>(@y1Hz0C
zvJw{(7*n1&lW2by%`c4LiLq)Nn!>S~S3*&LVFO0LG3D!UvQgfCl<lpIWHU`Waej^#
ze)C^}4xd`^+S?D9sS`ltUPj@qRc16gqKWmsB~4$7UqI&fHs*TbcKkHA8{(ucL!oah
zn*PgWLrtm~)rK;r{>}uNw;-3K4Oe2<$4vIDdki*3^11X3L#ph^;Ir{L_~TFjm9<$x
z{qFg2|NeRpGfOn7@IW~Z1W7SEZ4;@z;~)mU7(j)zt!VH-1uP!Sz&%U6iJp-KPOtoq
z?zh}&$ipR6_xBO39uTLR(Px39-iT?zyYcEl9Xu40jzfIjVZGXCn5MM}7TP_=@b^Jv
zSL6dUePltZ8|=u0i@j_^h%dgAy$!rtA?mjx%j#PDequROoY+iNrAj@8yj#QPp}{)8
zKCfv%F??4-mR84r$)rixpx{8QT=*F5)vaLGbr({_3&?~0iZsag1NM%r!VXbOTDvrc
zOo~p!u^LIRPuW23kM*-Jw(z-r&JV0AlA^h1tzqU=M^s`v*x<Xq2%8;oQQ;}*{H(|v
zS~$R}?;3$mq244yRgu*7Hi780f9y~oa+N}+fm{4LZo|x}G^<|;Gp?*A1y;uNIUfsX
z>Xl>XY)T;`H;%FMRG+iU`t8ZQG9Hd=G(pP88?5m)ZR+=+l`X~f<le+^a-HsDgd@8#
zZ~Y{C%C`xd*3W_9Ni!heQ#LHvx|NvBD8kZb@#Mg{=@_<iJuKqqr2S?RBu<gghB-<2
z&2$RQspo-l&?;i-5P(Zc4DjnEV|va&6)pSBVP*UUo}uMxDq7LbHC1(F?cJx*!MBds
zkZZxzoiS#dUOiwX!(DJ>Vk3VaQiFHj>;(zG0bKouA0vuosbzW_@3Mae;8RP|+^a!E
zKL@dEXU~SHZU-2>IGMcmoXHF<+`-(N=1T5q#p9N9Jt*va9qT3)lGBlgNzR9hnAuYg
zs-=sWc+tspeU}=3Q;%bg7-^6OuN1VqcnqiQ8i4zU`%!7Q(aIsvl>~2j2g=9i5$B{N
zBI2|h-Hcu_?s)-Ncs1P0O`sSb5fx&euosPcp0dkV%1{mQL@x976c7ywI*Vvh*SDrr
zIUxp_0u9vI9nYKp$R3|-)}V8!2?^ontoW(xP=%~xQ-ifIV8&CnBJl%C$X|eDPCQ??
zIe>3^t?6-3Z7OV_PxJf{UVrGv6P{sYb*~B?-}Ht_C|-d3{MDGQhu4@}nq_$R>L*-y
z?KlKas{lj4UF6r_a+KEifU*f|fiwRUf8Q%i(<B%34#f<@yVW6h*M2c+`@Idii==p>
zLyIwcV<oC;46&C@&2Z7FO;B;@9%~U8z)n;uW^4B4;LA=ux~9O1jW~1(V`EiWQ>{bw
z>EGwm)dBO+2fnauPBD%c3Q?a!hUDG9yQuXhhj_6T)ahvvE>p|IOrA6;UBjWACC&Br
zA;66JDM8B=FA{INluqT>XuGE>GVpyNc{=z3+rPd-!{83|fXnc-$c-B6+mgcjZZNc0
zn`nM|f^NO>pn)eblv~Oi`f(bK$A(~F%0Wo%nnc~F8q?{VD|kldEe<6urDIo>si>_k
z%F|KEY%{^kMlVQNWJq3}HYPd6(_pCke0@YTK>9yp)SJTrZL?7}7qv*>w&!4QIGB9D
z^@g42F`fJtm7-%){sB>ohj|{An8N2RI8Ke+XzR_ameGDR6_R9*M9G1s!aY29MUtLx
z@?l#SZ6R;JaY^k{84@2G!{lCXW6I}NGp=!m(P4c#=0rWi;S70T+<n=}+{tuFR|E~U
zzsV%)bh3)63Uv9Jc^GP)zz%7dFn-!6p{_H6Y`ka<63>+B4H0oNLH;j3YH=b$+;%o=
zMKv?0*%CceB|+}OExf#Wgneq*i&@8b{5frk`O~iOauXNAR4-5bH5yGm9BIPxaCsVG
z%-0ir0=lZC1C@@CGPcAR8Xb*5-`|d2kCLWmaxH21`#SV_nFxZ?T-uX(ADM^@)Y9n0
zzMF>;hW}yO!X9K)cVSM-C)UaB2I@KS*Qr}6s@^yW>i+Tgz4k4(415EXI}>T9aXH$^
z6oQP407>$e!v@it%**sA@b%wPT<&Itb{${YYu=~uZf6-+<k~Py%iV;0`4|ozEypU0
zLaR`#4R|m~nQEPC0#EM}rh9=h(eywz-{lsbELwzDr)<FT%Hv>u;wfVpRtCQ=B9QG*
zvGdXk$eW!3J$5NzBiU^A=0zc>eCKnpz0=4VkxeA`!(Kc&CPuV(NHR-p$H6M2gjrkl
z8iDoZANQdsZYfLty|U-WMh7(BcLOqK?<1D|Q_-UO0QM3&Y9DJu<4ZkZ@B9(8dSgPi
z>GQz(^g43hi?UIT1~>~}vBmrvE?#FPeW7<2+s5T-!+0X|fggja!U`CTnv1BTw-?t%
z&BprIu9y^i61Btk!AARTeA|ocSz8r2tL}i8wyYtmM;mdNImX@Hehq7y8t{feKjUXr
z!yLME180ngW7e;5Y+JxzALIteSTABf>YDS<MHxa$hXc&lUrui%4C3`1LvqtalD2p_
z)8}`m5VfAuMAIaYypgOy4M#;fCT`FEvD-tcBfjH|`ghnc`$&CD>Lo~*5Fk}ijzpi!
z&nek+XcK#a$+=O>*D^R_{dx@|by<Ss@APDjW>2J3-}|H6!jte}t}ZbTT2J;2tRvNb
z4zPFM6Nr5N7<<Puu>PqwZQQCrN3MLrrE@2Me^D)~a6$u@RQzPYI38V=9cTM=^+@Aw
z7xsPkB_=zr6_&<rpbti`;orS1{xs2~rg}@*9HHy*%`Js}^Wz2v{aFoPT!d*<&vjg0
z9S3%)WjLlSMFy|M<1l}mL-VJQ9oI~${yHO?blekN?wmlbu?Kc<PG;km6>$Y;zJa4D
z;>62Yh`zm-Mz*EqqOT>tUXqzeW0R+0=vx5M9o<Z5lQ{i!K%V|Cd5cu(J;UD0<4xax
zj`2IEK-RBO#|E1rUbP<sYd;^rS#r1F;LKCx<G+ijwrdl_s^$=J@j&va`yG~NZGz69
zA5a06NSdt{$vY$qri<S(ak8PfTVIKY)u>QAvsvgRWks?RmM}x4&uWH<2I=R`;4OGu
zhKe_);N7(!A+co~^IMj%W5vywu}+l)))q2;+iTF!(;qUyoYsuF6H6@)tQ0;=Hn?Xo
z7C&aA3|ogEG8_o;6NIKoJ?ydUlW^!<5^h;>3}=3p!HVZkFrH1r7ny?e!Nu=bB<)EK
z@9W2`RkKN_zB{AYe3mJcKgq;}ABNV~MUeAqI;p=x8MUj@bi2z`y1J&0ovyu<+z67O
z#)~4*wfY#|RC`X`&M2VXZXIedC6jC|;m{|OCQ`vjU8>JNgLnC6AAIA9(Z=i}(AGQv
zV~#;2G_D&bUB8A0T}0{q`p>BP*NG+y??v-=Lq3lqPVT9BfWrP(X4Lczrlu^xl7oe?
z=)pJk<#jF?rKDqFh97(lGbD4;b%;;wSM20m#>JPP;?{GIAh1S~HcP}ZK@HM0S3sMT
zU29>bNdRB}e4A~!P=mdiswDbZCp0A0qS1K)OqsccRHUer%lo+8+>^qztT>ivt$M}i
z4?56mISa_t)ygDXG83iOX%IQoCVx)O1x|a#RW02#l**h$?@CU@9hGx%vQQUeAJxvp
z#~!WUQy@pr1*eiy{nyy7=fVsft+LV_Phqd@ktgdCwnF~iY*4GPAj6qApfrkN%lrT$
zB6|Y;UjOA*+PuN%v&-2PiYFm$ss#-@a}5XNwlPh~cQ7()HoWCCT1@>G_LxBm*iA8`
zYu;PZS^sn~BF_tNq%4Bk3}dP(IDl1N*D)(rfxIf@OZAs`<EmOS#{U(e?WZ1Mhx27-
zc*Ybo*?t;doT$LzM^U))M>?jo+tA(o{pxvbJ(_1#&c+{@OS?J$&@^-p>GS)IZI2`I
z?%*9(gbw1!q+3j_ff_B-;9=+LIrLJmGsMVRF*9A<Nq1Tz8{c2ex*3*po0a+6!9hvF
z^_WE3GL!Mo)m&J;DUGE4ivyYTk9f{Wn21;&#8N3mYPIGMTW9r_{SJ$9IpxyF-=5-}
zQV&eOp+nzz-T)WxbadG$MyBn*js5WsxZX>GM7d?6)f_%&bAKv*lC_wg2n)yi0S=&Q
z#-Vh-IrFqYjLPNagNkerm`u&arxA%59k~*hy7A8(8Q0@O%GX4t$g<9#SK_`O`|yQq
zE6#oQ13nnZ5W{u@zNTB59w4TqU|t-G<Z+n?Ha9WB%n7N3J4w*1#xtqYC<*U@zHu9N
zU&jb@gnt&YQnwWKU;5Ja)b|+IoQrT+k7`O)FqL0U<7-P%Vw|N+c)Jvd_+%c4@*d+S
zl}?n)k$`erFR)&_n52AMORCQ6GD%t^cu>lYg7|6XC^rvAVKtmPBtbc0sjPomHBrbv
zi`uoD@s61@>DsS?)mdEdtuUrKA$lOAypb)J7p8A-%Fx0M_i^Ba5WGI}8NJ<N2{HNr
zD?ItU{T?-1;LEaq(uC;lGffy>8Hj&AW|1>87W|qr8{f;VBxk&XP${OFYxB*VuVE_3
z>C063oW2d+5`P$dp2<)<9d#6({gD^8d;*GVdXvf1Mw!sbPLRfr&%Ls)Fkm(nvNpV6
z$QuDVfgA<<S3OK)Oel0{eZ(yj_?S+e1dTs@9c_Z^AbEK^rY%0s^^t$h$B8b$+l-gY
z$=B)ZSYsR+3Vn%loVMbON_D!!Pzq+0y~Oxew~_7D0%mw3>n0*W2VY$V+qFD+t<k~A
zNU4*&U^PtIIRqkWKESMs2qOOc8fY%fCe0etS*x8X&>UZf5$A1S-I4d8|8*rvY?z3@
zzHdO^DRYUR^9Hzh^d@#J6{C-T9RZ)LY?Nd~=*}(nxN+Y`sGg@l?dyI+(`X2jcj+dc
zls*H=m&{1I?_}%}tY9OrN0Z?LWq67arJc?d7<>9Lbj<GLJ+r%x$E+;L$TUZc+Gqk|
zPud_pq`3aVJOzl}e}+-w$KC#dY8baLBBU~p*e_~@o!tR!;a6RDhHViv8ZSft1u}Ht
z<2ewP7Gq!E`@<x?oJeIlg9s<_BvZA0H$2eo!4f_$m|_$~`bFpdABxU9AjkKM!)@)Y
zp&>~n6-D!X&MirlmQ+?~AR~mtH!GDSNn2YbN>-Aj_c^x|Ns>@hNR-M-L`d>`et*@U
z^}f%z&pDs#I;a+KjK8;A4w+NZ7^li*eV4rAn9K!abI$>y8yAE=a}lDrb4cUAzc?}+
zjcrF%>AYFF=y)^@FLnm7k}#jV;bx|3r=Q`3==ZQ@uo6?gm5^skmqE(fLbl|<4ea<8
zh|^M~DErc#mP_Wb^uuJbDAFFBzB^IllZS}PvxP*4W8+;fN|9_?SrjpwN$q2H6VEYA
ze0%XaTI>D8<-RlV%ez{vTo-^{KdtEINba03If>R77T9X_O(Wr_Lb14fD~jI~rp>ty
zq@@Y4Brc1npL2sX6CXqH#6Cvzh$QT6$R^)X7V@*+%AiD{J{YT<z~MA84q*HMeq1hI
z^-Kc!oD<Geu6~3n&T05^u`zu-V<Pt5EXHmL6|SytK$-l@XskaGYdX%7O!IZb_~%L3
z#^uj6{<MQY%@PvPwS(7tB%b-CWdc1;*-YZ}$=Ee!hxN0x=`<q?>M;E_QyghY>T|3~
z@4tCmX6X`Ma8<;iYcr@o&of&U!ya`15J~Q4i_)D-dQl}j2OZpIAhvGB;g&{R+Sv*{
z_Vb7s*XujFvxc$RM@WF}6!PC%VQRNKi|jRhi=C2laaSzI3(C%bu{0&JGvx%w(-rcU
z-$2;g`4%uZ9vhd|VsE)Cso7_QYxnP_SGy#TEqTp$MqPpK^BR2L*%MJEY9YDOwGxt|
zzk^0i2CPi~2u&fnL^e-}2&~_Q0+APZy(L%hQ~e<3FZ5xTEDa$-k}FW?V=4Gvzrd88
z2>=tV3!pNXMSk0K;GnJsb@>;E(a$)ifaW;NSf0%~*vm3Q>FVfcmCQb=6r%b&rx1yn
zLUi492emb-*f*j&$mSkJo@hR9({seAIA7Ye#E_1gO~b99oXA}+f1t_D%&pU!8T*v;
z+&t$X5!%rRbwkf#%ukSZDPBP;yNf`LE~t0sqYQDwvX&#vn#2k8mcSu$MP&f<DxTpZ
zQ8(25r$ElVm`+0ZGSv4uSKq(#Bke*~bnC49aHum81$Jw*$Fwf7OA9RNomPeh)WqO|
z&4sXEdI~MuS%i82#lk>s59|yev@^F3?|z(uvxZE0&W1m6&UszhVdG4$5jA2oeGA4{
z*npk85V4DYRlR&|D-5>XLdD=6SYm#av2Jz4UCjZc(AX0^$Pu=pP>Xm}l;c3X5?#1B
z6KbDvIYHq}_}F4f{X)OqT=zEvs(CHAzhe$lYpVgFd)8p}?gzNZA_J}pUc`%y+}vAP
z2CQ{w5szypn2omSIP~xlKeJtfdeo_sh<z#5E8m|0^{AhiyK){CzcB~8mHbi9wGKaU
z=Y>ixANpa#Ji5q6g%LK4BJ<TaXUBmAtnfJl3fsnTl%EF@gACD6Uj~9dOVBUl7K)b%
z(`Quz<oTL>vS-z1uGjt$@?UlG&bcqeibFvtkhkaN22(X!zlZA&w0z}nmhWL>S5F}M
zCOc_AV?fq>pFz8)(HO8)5R$H%gTD505?ipEH}|j@9Xln-X4n)Gfw0GTd`y_s`Zhu`
zKNN%Qk!d_-L%rtw#Io_FI57Jpei~?mvhO0yKDmB4c-a~UB4^N$)?kABkCUZ679@M(
zXI!;dn+VVQ1WT-EKy1D=)%3my4rP*5a_K3OxMwvzxac_;6{mr(Y9zC4VmamyNzko6
z99#ZbpDfvMf{_e$L3OpejLOPx*5T<fj`QxRzT2UM>s)L}fo?o(mgiz$R<B^>KV_P6
zU=9g>aFShg&X^hWK8<r4r%{dH-sDnN7szh%XY(-|^DHOO<=2leIvKywr9y#rJ}P12
z=5bv0_F2T!?huGKnXtyQd)b$1Yq0W(GF%R3+0b1^<VoCLto?kFmI;nxh)*wEu$Q6V
zQuXOk?q|EC75<fwq7RggWAj@s8zURTeNRzD>P9K7y&8jIB{66?egMl~SkRaAA2J_q
zD`HvnZ1$4pA=I;2gc-?m@!%6-6fhjM>ah3)K27s*-JBfseQ%HHiji#1i8Z`5?^-Mu
zP^P=hZbH%EEbKioo7p(ElH`en6RRhW&>(0tPFyHLtBej}O_&@?eln%`(;9JX+FVjS
znMW<(A0_YCM=|2VYjKnQO!mXg0caLaBWBZEAzwrQhBZ^Ttjl-i+Pv%d)>@RxrK{3-
z^N+Z6I2N2L`yl<*9->&C$Xt;PC7&+1Fo~yzc)yY?*s%+;U~u>bM$gXVNsVZbwCWCc
z*vfrITC2FseIGcOOa#{8IoOS@X9vXdaZ#l)@7%>9h8+!t=W3oLDEJWLRq9WT^KUXi
z%x%VmCr-P0$C&u{6Ttd9k3Jreqpx%X$>V{4jKSr}v^n@BL?4kLe;SUG;}LHmO!FS5
zw?4&aQ7J}m&Jpa|9YpftV%S4YM^IvcDssaL5)xrVx)tW~#><9bzmz$3@0mcp78c`1
z2U}|2?T?iUC(tLs4#a2qTUP$TPyW^m-e9<k;I)M(KwhE>|8aA==-$JO(Y&+F*w&wT
zYQF*fz|6tp+suhd)fD<dIGYujD@VI^f7~=0e}!#n$A~pv1DQ=eB)Sf9s&XZ+%zT1Q
z_P?-Umll&>Usdg!+kqvWi?R6SS5&;Uiv%s~!l`SngHGl;cKGa1*jY6L{S6P`@imi)
z$eDijM~n^%sGYS@E7YNjKQ!YBJtJfu<de&<`*HK|HDva6^QKu?QtRTG#965lISmCF
zyfO(`jw_p$9m>f57A6|X;qc|*B;q9S2(JA-%vR+5fuRMl<VWu<W<X^TxvUj~Zo)T7
zzGox9=3hTnWLLtv)BDJ*S=;#uXR~m6(F@#?-U7>lUSgcNA<0n6X4H;Kp{ea8qH@ra
zas0joXQ#|0@i7z0v)D$A&&^=xOA3Nt*dmhu$Q+e8p7p{0Dd2zcC4On~gfXL9EG#SI
zxRxFGW>+H)B}g;M+l_IV#e8fK9%98>^~i_4LNw0%F#2|^B>I*5xND0RH9Z*$ldN4}
zQHKK!w`|7RDcU6Ws|5O`|G>lAXVJsQiH^1Dkkj@)B*LPYiCOD}`*#0;IdK;Bz>_Rc
zDjo!94UX^I(}U~(j^lNnJN4DDgQ%uP41W}XQ|se#!}&*yz>!>D_}CK^dYcF>1#SH6
z(}bws#b^wB;R6d^a_{Dsji@uThCP$62U2Vd{^;je0JUOldHw?LB-q24XcRfNpcRyN
zK4Zdf9z`wCB{e&*fManG(Ukv&VXri3+@lj@`8;hhRq{ULr6fb#cMH+wtEHhmSe!0(
z>4HX=Y{ox$9Z`J2^#w92Q8_FT?Ox2VUG(-3b4zM69sS}$_s`d23lh0m=D7%d%`fEt
zS~Y`A&+f!6r8B^C(3D1u3EK(_TSLRVXs{~`A}ewc<6m*}l5;t1g{mSQ-DgW@bbZ4j
zeHk+6zfNQ$tI+r=*WUvJ?07hZ<|oLK*r!6wLVg>{o{WT)o4eV8Cz2$)8nB7$xnwCi
zLr@SOeYVt~{_|ZpW_^J5@?C><ikdi9E=nIfl7tbvPPiR7fjarFfW5gLFmfmn7Dx$F
z{wy&vd#615IX!`vyii7d!C7W;(@*pfGNC&8QjB|7BEH!ajduer=<&@9>E};cWUpog
zle*S}WAF=5Y#ATe2{FVjY%P|~Z-<Rp22A&sKkN+s=kUaD1Vp(r)VbHXc*)uU&Lkd3
z)w@aHd!?WC%d;Un8K7Mf^|0I_5-*>yqw%+nfrx)T6aL$d`M%5(<j7s9UcVT#U2nqE
z-Rt0|bSc(F9tX!4*<?Nc6uBlZO26lO5XtAJw3Bmz92?dk|E{kmte!iK5d6w^Zwur}
z<o!UVVFxk;&rwB0nMj0AgQf2?u)J<BDN2ukJ(_L!*|ZC_MgwU=-6k&USjyZcBM`&&
zFO|8TvRkh-DSsnM>rMBNME!TX6XN;cc2Sr_@Aqf&@+qd9ZNegpkJyh7F}TQ)mUHfr
zmDS6k^xQlgp681ZQ&m~h)?Ij@`YJ>xiZknv2os}2RiODjlzB-y*s5k->KyVL3Lhln
zX9W*3HL)1aXPJ{30VhBy=Llw$Q+}*dC?5QBj$Hiv63ZHH;$&|A<Qw`BH<tck6+HtX
zJoO5OjX%bGI|aH*Sq*IRjEG|wcN!cS<d>R>Qr2IRHXV74hK6HU<d}s2jpl;IQYVts
zmWEXeff{**qwDtB)UCD&z8*eCPMDr&Bm4z1;*2bn=W?G_lP6MvzB-gyB1^JjgYYmr
zigBZJ=t}co;$KfW9(Muh8aTxih$P|ue{*Pp@-#B$oXma@n+R`JuH$RrdGxTg5GkEN
zAjjMo%ACD;i?*L2TH7S4uhVT%UVV=jlVE_ZL8hqfYC-Kgcj1O1GwSuDiS=3LkK+@h
zXmtY*?M$Nhc6XxqKl`|xp?Ep=M*rk;7pjm`8_V2~_DAEaLqIRfpj*Zy8e%9yqnqEb
zTSnxG@<#(^yn%Dact)eHZy&bpT0u9w(I$$=w&Nk!!(?)P4(hxw<0-7|K>=Zx>JN>^
zI8U(==d}vqoiokMs{guhg^n$qe_NFfp%(e(VoEd?>tLXnBPOLxr@GlW@G+$aV`5HX
z>wl>@DiT5VdO9-IN8Y01yN94Ba14cdtjL}#0<`s5H$H4OCNa+hsJibd{Bul<R=JIX
zg|$67%k`T_)3{8L@FJ2T@EQ+Cn$qpjW;8TxG7A4fcy}U{q!*jw2Gde(e_77!dT#@^
zVqY;<YZ&)`S;H*ym;k#cEasS~Gbo@}U^DY#8ff-?W>t^AW2J8W$4XC6!}BS3VMm=H
z`b|;+yM-J(!_5Kv<Frs@O%3|pn@R_o1DVGh*MIqY469jYNNs1_Lw%`Zs2F8UKZHC(
zrwvaqgk6MZdxhzm<I}<K+Y4U4++y-DF9)Vg>A`c;WQcr~2;E-nfhDD<*|XtBxI5DX
zHcmI86BbXxi1B@VeTPg`w>BW^NB>{b_rmd*F*YS&EtDVH3YmV%(C*FRhBaxJw`?1+
zKD-Y#nOJ0Ob?CR4wY1}wHl6)W6Vk(NneK_z{1X*_u=~&h-pejYG!wmq^&@ZbEyvA$
z_##OAx@S_WuLp1fQ6Z(>l_1XXf~zOk6S<z{%xYH|nqx7WZE#9uYn*nId%vgCTQTn#
zXRi+2;&YHl=iNo-m^xi^?;e(#7PIpQFJaX*9Z*so07ctJXm2tRAFlq&x)lvz_4ga>
zJT(=x6$?ej?}~J=DGm~P4q&TF6H~Q8AFVBJa=8m1MkXQ<97B9TTJH|l6#l}C0=3Ml
zhk@|wvl4xB<}9q_dO6KIq;NSihmyL_n0~Jm=3QP*&ZW%7%OeMf+GJt2;r%s^xtl=z
zP6^Q+{QLN;;UlUNX}Zu(fr!Mb!-fT)Q02H8M*RywhqQ_C;{H2m5Yk|lzI7wgkqIo@
zWkOe;GNc}V&*Ikn0rW0Df>~}#v}(RGl-+uTMz0gFsKpQVWZ#GAX{F5Nz)t4v**Caz
zIYU{QNam1M0VM7{&$GXG3*s-H#kePjv1P7Ab^d7u-m`W!5)dy#otHkwCe9^r;CBw$
zbNwF9YB*CJSDg*-bbg>@P!C>iX=6jUb6oQ?5vnV{i=@8&2^nW3(WbZ-Jr-wS!Ow5_
zV<jQ`t}h_gu?I2G#t`kETp*)M!qGlp6x+@^Q@SL9Nk8R<Av4Qx!pbu2a226!dkGs_
zbr}zZZy=%H@|c9s-xy^iO|`Sz@!gl((7${d{hU6Nz6`z$&UchBbXO&_uSSSecu10e
z+9Ob!V?pMgivg8_0rdS~K-bQ`gkdAQp>ek~{#EY9b9<G*bG1FO^N+`E^`W>c>>Q3z
zTf$!d$(VfN;Y&$<c4W$Jw!X)dT8<yXMBgarYgVG+T+i8~cPcx+fa~LI>SL0>SK;>C
zi{Y?^AzYqz0z(aF(LSX$H2v^FOgV2)YvhyBxx|(j$_cZ+j1rmpaxzM|eZUY#lBlkI
zhSduc@R#s!{*r6gu|3fml9!C&0<~t;KdVd~efE*^5I(jY?`0Fj4&sD0PVh5)F`oN1
zi<m8)KyNo3A(eePOjzhgq#O@_|HU!vcG<^7NC?tFiv;vFSV10Hap&!a<2Wkt0BSCs
zhAn%ikXMN_QPX_}$zviI`4{_Hosdit@M<@V*F6UF@>q0qUPbL^)S^nv9BS$)h&j{h
z*u(ipFkyci$J&a(@|{8`|FQ&Ly<diOzYpCn83Xm}CeTXme{i@bPItXeK%Sx*QNFy1
zE#I*iRn{K?-@Yf{u=6As*1y6*Z62BZX$Gl>cWnJ7RcaAh3|GGoGRB2|sJ+>Qda0ae
zx9N$HDfMbJf7Lnu_R0TY)$M~Yc6tmq<tcJ?%RDk|pc`U^<>;kCb*!1&0O#WWgXNxs
z{J(xk1~%4X;vFFpW-x&rxEF&V;wI?hl*!l9?1jlSz3`+zfD~<ZClzk4_?)Y2eYYJY
zg=WXtU1>t78(0AU$x2ZT{&f7YC<8Z_DiN>6sr<d8&RAmj7;85#r)N{|!oH_<L^z#L
zfw3Yi?6n~d>F(s+7C*A$+5}o<7r@S16oEr3NBFxIC{lq&B}%1a!A$%wcK>&XI3zs8
z=*K}wGCb%6e@*K7R+xHxT)^J33Wk;4v&nJGdXT-@gLuUmMrutd<?rU6#p}_#v5}GA
z`W@^(9{~|{XVPi&5tVx~A^hSo80gmkbEBuYEVm6F>?*^|+1(h&&D480uO|kfJ@~8N
zh<&-^CWhN2;J2@}xN?;pb?S*_@?s6(XxxABC8iLyP8Wi|gfi9FZ3Nu`2PzhOlua<m
zWgnZ@W2ROSIXYVj9PayI8Jk7UYMsRdFLm<i*8`?3zKS_HM+0V~2?S?GfX9h{*n27o
z!@4iyeeLfUk#nK?^>lf9X~t$uI^;peXKArEHaVELWD@DxtVznW?lOig>TF1$D6`RT
zGHsL%hY-kRogOu!+=&uwa=y$Mi(8T2s#AC<(+`ZmohPvC4Zl}vjIEKAMc>C=P;uFT
zj1)9tcZngdWW6XgT^|Jz$0SG>n9)~jRcZ5jaWHJW422QbVDAY#7NrH~6ZO^fOH&*%
zDz{_q^6#+y+rBfJV~cU)Nkx=P<Lc0*W<&|z!PE)cA+K8((nj>*!-|XeOg|Ew`hK!}
z@6~wXtutNXItG%SQ)uS&85pw2kx+wU@J1mPDm%IFw?zlFV@1%pZwg8dJwyJ{<ESUw
z1r`VQ<Db<!Sk46DoXTFlmq`$OH<*Xb#z(lksupeT`iBX#nz3p894<?0M73Z^B6BX6
zSvGq<whEkw1KWg%p&mnBtL8!3OMUvXFCWX7JYo-aDdHsgVa!_c537GHA?j1D$nd3W
z&^(xh`TN9q`TxRTVCW2^bJvKpDyQ>@x^?kFr#@L1FF`clYm(Zwd04JH0kXdxMU5#k
zR54kXs7JaHeeQa?IH3>s2P+W6gC_I~5vG&e5};vO2V+}Mi;10+>9XWHR#>E#jrnCx
zj^2tOEN8uW`TZP@4Oye<seLHfZ~*#kJDH8?6VdXcF_qlX0!71SR8RaV(?)h-`m#Z6
z=nk*mbK^4@I=GQEE<3e<<$aw0BM6gM3D9_(!+?TM*@vx8xcrI|Q`nFQQ)^4fny0(j
zf~u)>Wo9P#_r;-_?<e%xJeAb9P3I>JYfyvlL%7?#2UTn$*fuWfYtDUdRvCwI<<>lq
zkJO=l@4~^jyN8=2Mj_++5)3=6Fy!Q2e!STS*y=e4ovti^E3XH!{F)$UMG3QBP0!e|
zhMSCB^bSZ7Z^ut<swCg^9B%FO!VNKM@bmI4+S69UMX<Qs{`q-$=&&RWwa}qs8K&ex
zn<cT2QU(={=hL)*z{I^VpmVshfqhLA&i;6b%e~x1TfaHLDed4qmx%}sW3WBuC;zF6
zH~wx8hM&ltlSjI%eYbzcp;bO~u;4U4Zb&4rKJLdQ)+Ut6T8>}b_2}+W&S7xYj>axh
zqt3QG9M~Ys7!Ny81^qd2@XKpfVub<;7Vc!F_j5H)at(gCvJ-_f+rY<C8Co)a@spLb
z;Xw=^kM-{%<_li1v9^-r<USjW$lA#?H=JhdE;&?tlyvbnOud0;R!UO00CDEmRSS~5
z;vj~l*rH=Ocb51Njq>}HXxWn{rtjQDdUpmltMj@?CT<X>lyj|#{TYO51sloD?heS*
zaKW|VL6~j81MenjtZNdZcaqLCirgIT+*&dE&s~D%ar4S^A1bh?r4-Ln8!qpb$5u$)
z$AkGd&=v>S?x8h2%e2o-S@Z%z`L8ffjpNP@9B3cs>ktUdzbSh32ip70AyVJn$$9O1
zHj(`1tI3|mG{+d?y~32JY*52}aX&b=cL<hG9puG4vZQuXU-6~=E@F<yO9<B(M*VVA
z60NU6)Myrh<4H_m+hL1?EGg9P1J6(WXmv>jC0|`;?}Y7RpK7lri4Wab+h=D<R^J^I
zyCXxpMc?vcd^pZh$Asv}OEXsv=n{71FME4=Jh5{(rNT%35mYMK-raZ6wLTs0gwG}3
zSB`>Pe>$U=^9ZH49)gOS5x6L~oxknPOsZ`yOMOD>ajjVlFbe9#JH~+AT6qEu8w)Yx
zcMU7hBg%U&dJp{a#L?`R0&Qv9!w)_050+VPVa`U5Uo&mTRFy6iX^jG*6elXJvxl5G
zBS<~ko3N@#73R8%vC>V8uu|p(t_}Kzc~b?c^^LE5odbHjA<p?))7HTZYOck7&z9pK
z{R%R8UX<nwI>6QZ)6BT{B-9m@A^VNXQT+ZZBCt-#Mme$w%`col)z)lCpI$|(lZ$Kx
zO$12ax-A%)Zb66bL}|u_1k&$zA2+y8#N~bMZ2sF8ChxvIk$ut3w%crFA~z@C+rJla
zZ~j*tDB8>4<Z4e$-b{pxdnKva(`@pZb6cg>>NDYW3N-G56cI8}p)c#MLet(foD$Z=
zo|aZ52GcGxhEm!jHS!hS_s9T&2OIhNTn=TzNE$lqGoU^rn@R3UVX9YzxN)s1ndPoQ
zhf^%6t(qP7ek~?XI44i`Mg}(dCn8^M4w_m<GOn_f;I_jS+gI@*Zb=`OjwImJPu8^2
zN)~-KZG?Ft?TpEjc69$6hhx!tFt8*TzU(N2^U;T)>}U^va`0?0D2ZaG2tI+{(KskF
zdkx)>uadR>`q=bi3MT$ZVC@`T;mKxY#>bz_d+#~Uo^DYB9qB1-UqLJ5p;yAQc$|Zw
zCfe-w55@SYeE}^`N+NG%_Y!7w4->X#8Wt>wAg#8q@qY9Z_QPUtk{wgVxOa=thV94M
zlpRL!VFDlX&oZRUc^_%2P$Am7`|<9h&E&(s0m$X<QGww<_}0}7?_NklxnvgzdfX3T
zV#mm=^|iQr-#zxp41fHm^@I^PeFO{Nm9fE63m7*G6=ENJ5A<|B=&$x-c7vuK`tFZq
zKRM`=)-AfMf$1k4%uz=D_qoVhlZDQgZlGJMCy{?LgXS%;CDyC|@IA*SQvaDv;5B&@
z?lhZ^0uP^I^1D7{_FQ0`ZYWau(n0pj?lw%_<p9Ke9<01@fF03O#*m1c=zskZ4CM-7
zA7+td$A7cqb)J+F7DAI5JBeTJ6cVs=4mGd8%#NpY;+)nGY}o5J*vpS#*IE96eoX<A
z@bLtZ7eCCdyE2i4PZyxQ*QU{Y%hPC=od%Z`dbs{=4ez-zKv~{g&S5o%@}8Nv(QFN)
zT5rW)scK3W=3avp?&2hV4v(C4v7)b*orMfTgv)|4?E9N$v_{93tY1)%)&F_ZnA|A#
znebuQ+!@Jz9_FmxhHXf7vlxFNIdbROVG`qZ2&5M@FarY#cv&t2``#up3p4LwgxCZ0
zTRMl%s!(Twe+)9Uq2*Zh_b*%z9%r3ge!z|8hlu@!U5u0K5*S$8&;KRS$A(lC@l~A=
z`HvUChWUCl5hQpq7mCnpOqZM;Nntt;j^p`>y?A3pk97G)alHCT2$=X2iBJugoDsqN
z#~IjA77UW2)5y@0Sk_umnY`~>gG#43Hv3*2Bcd~do_IHbg0cu%-t(CEutI{0pYq4P
zUSaT{R;=`b5?o$2oi5*!!uF*vBSQv*?08TEI{geMcDwIXcbh)PD!EHsO=OE~(Q&j*
z{e>Fd3*cD5RjkVzhWwhPAjJ7-^b1ciqPB-I{;dz|eqR;WIEd5T%~9z0M}}tfULqxd
zKJX*kkaXwuphu@BR<YNyRpuf-v0ee@-rI=5GHH77$2_#p(WCDj-EqTiF~)X|9{Kb*
zm-V{n4-sdTs7s$OZA`5}+wq53FKEnPNfn^(vm{xvFA|bgCE<|+B_JfEL04YS=K7L3
zXqU2xy+80D$RsO5$cGbTN1`3gONv8Fzi@aFH;M&6%Hb2&FP2L<#aK#ke6Y+?{`qSf
zG{V!VI`qygx^3}SMn}1iy>l`NQr+8dxojCat$o2PThR_@&u@liV|#kTb`>!m&Lolv
zrkE^R#YWf2La^%`^7r#yFb^-otgc~bdS8m~k6$1sGiN|i-bJRn{~BMmZ7yE*_o04e
zm(gQU4u}u6Fw?^n$n+`FG|tC~v{|XsB1Ie8tys=(k9xv-$k~!tGqX`#(3tAD?gCEg
zho()xne~UX80#^2BF;I*x+VVN#liz<9DJ2IeO3^gR#>3R?h$6q!)?T2&VQIwWlDZ%
zI1|xTjgWKwDLPy+L9NGvbU0HRjz511!XI6T+ut1Sy_AlFX13@nunVl~l&SVO=Nh__
z4@HNB>E~6aQJbq5mnrk8)dxA+FKI|at!5CeqfZL07r=IbrKIzG1io9G%t{R0M)lEf
zG%Qdj2VGy|FT2OkD0rSF6PJ*1!DH}ay%xyVs*_Zm5VUGmB$+OiXmm@8v_x>{_>xTK
zulxwedH%u;GaF&y`WI-^*ubt?=F76{<j8>0F$jCi<u8r@;I42ns?ac*?j6r#o)otN
zbD@u+QR|r1$8O+%<6B@P=oE@B7pE?6MIi6PWlZZRuHX6-YL^IM-((45LJKijIG?F0
z8?$X1%OZj*#aNL210^<SVNux)P&<`~rrbMsK-!fR$q2!~O^q;UYm0{8YJoqwA5Pg#
zqI>i^aoN;xaz@}XsR=8_R=0a-P<|eLANb>3f$NAef;hvu2#b~*Q5BDS(96G$<hcZ`
zQIX?$Ic_FXF88uy8cjHBR2{<#EAedM5g7V3iA|nu4Z41uGpJt_f^P#|vQG$H<@e(h
z+af05=m7fcnM9Rx-KlZX9PkO!XYPz_Cr7{Bz|tUT{5re}_a_{MaK9JW>dVa+uZYr?
zDG2e=Z_#ar3Y|39nCdx%A`97Y?!E@`wOzy7ZW5=kT$!4^cm!)q9-+^H&4d|#022mV
zP;BZlq6YnpUBFkgv%JguCbJRCuFBByOX_rOSQ55|<szf)%BZL91IsOfRB=?CZmw>}
zEa3zaSDysh>2o=MNfeayJz^?S9m$idCA6(`F1#BDXp?>i=Wdt~HCrw6VVeiBY1bgD
zM>!Ataz9>KlnnDT_%r_J(E|ffljywem$;#51JNJkTp<%FrgpUQEInUweZh59NAQ$w
znNTacp*050E&CYLLwYpw9hVE-b`Vc)PJr47S-R0>J5#H79q;IJeZKc6$<T{(D4k_T
z)+vh8)$=3qkBT|f5SonQYGXKiy&8#*h{8tOKS*a!q<fT?<I$Hww6((nAId#ujDBkp
zfo55oJ53^N^HO=*Vjs+eK0HFamfvUPpMT{i3tV7_Y&WxO=lQdLPuzxo&-=0XMLY;>
z=)n0euChlqO(%8D3fL(+mu~nk7b}f~Az;f?5KFYC22+J78@K{?N?+&ug{Sg&4XcuR
z^9+=bK-^CBsLoGKn3Syn2e}+mg8wg8{K*t1OUVJX9Q(jB!Gw<PeGB<pr?8J&8&PDz
zE2bt@ny<E6mi<%o6$=k3(Eyggr57(?@bGK)A?{^s1q5lw<S!6Ea*FUW<!Q$8v#fux
z3V9gji(4K^Qs<U=ba};jEb5oR?g}lo@K6A(vrR;!nKop1a2|R+(q?o7Utqq50k3V@
zFm|dPCDotJAdh#HY*U?2hlY#zhv!+rz%5l=-C#y!oMPF7!TwZXs}B8C=|r2%j2NS1
zNqA%0UAUY%jJ<E3Gs>K&VEzXwnzQT?o-AEN1egECkJc5iaxp=(^Dn{p>t3>YlRW)&
zXcTMCIq`ozEWk6?x7d|sTEu;Q5<ZH#gcl!=W5@v|DutY5tSkvUf?KL}%(HAGe|=#*
zB2M#edhZ40Bl6UFeIhy7-vMF27m;rNM82zF1o=E&lwQ79$!hN8(M6$wtmEB5RCE!e
zFRxhB*}V%uRX7$y=nk^^%@pd`y#QllA49BF9o9^_i{AqUXzJ@eT%|di2yoYR`&V^{
z-4M^aeAAhF&a$Dt7pj=p|NO{BXNK$)SU|1&xxD4274!)E9P(CXuzlSI@TcG@rnd26
zU~eH*hI7uBmZOY*y#^e)FM*>ecWj=!a}KZVPZ&n{J@a&yDE=#H#n8FlC}o(!%;Ef<
zs$b?ZAN(~?T>TZpzIlaNeK+C2`UG6|B9Qn7ic&Pl$8}wb)W>lWNbU56)ham{yGWIp
z9N<FPf(CX^8jp%NasIWMiF6AY!~>Q8V9Dlx=<rmLb0gk@496I@G{h1YNj<bJGtA@m
zV`M{TStD$;8(<??39{{@1iWrl19>jjud-H=iMK7sV*<_`FXoAIVg_vcG$A5CJ&oru
zKZm4pIfK#Z;h>!yjEUVPJXT;Dqc*AvE_QwFe_o3*a=Q!|)4%MmH~ZNA7qsXspWFNy
z&*Px#z-81K8b;Nvhw-4qC5(7z&URrB?0e?OeRrd%<HhIgENx+L{Yl5|5AL&4oTDMM
z*MUmjO8`6Fy}Y<5-2M#P`4s-f5Qh{05l+*NNAxU60|*e}nN1Y`oWZW+Z4jHB1>YxL
zWmhW1WBsx%oUi*H-Y6F*3-VfV?>&3m{EKtpL@op4t6$-w;8V1==5s!ix%?8P2+Z5{
zjlCFX52F*$fnRzqe3iVzobh!fiC=H?M6I`>|2rc(^WkbZo*Kz{h1U}W&I_@jIf^NL
ztU+SZ-hu6-Vx;sEiU_;%XWW?x0t?Nl{Bwr$Z?4C~jwa-ncp$VtO=pZscEUQFm1N@u
zCpPX?G)8cwQ=BvBwyD|#ZgJiq>^GU@3OeAfA<kiC@w)ozo9RSaX#^wMt8N+&$#Pzy
z-6(g)i!QW0jtAOm(Dm*Ys3>ryF&`6A%;Oc?&Mf2i>wg6Kyh@n0{v>G)dWYLnLNMv-
z8JrwZ%E(;120KQMV6>ztG3XIO@9iH^`>G;XhN!R&8yj%wJ|TQ^5q3n$pl?n$nq|q*
zU-I%~^D}A8%lil4&YOXR>H$;=Oobtr*Ld6FB7EN$&-A>r;;#1u^d49bo+oFLhC#yh
zBoQxVH?oz(Ct!!`ZgO46g?dQo<KnipB;fTeTx)tCcCVO;@_wfL$XB_Tp^(G(kO^XV
zyC!fuQtmTjdw#&S2X~=H$CO$3R}ov+m6O<DhJE3mj$>MX(L79xsz1v?xngVF7*@fE
zy^tWkrxl@Elr(*EeJ+`++=!vZ9Pe3WO#N1cF{w*e!mQt2IMZH{n0J3b+Z-D@%U%}5
zx!mUL`MNM;;|lElIEj{|Mzc<RGB|!I3C*?7K-=8|nEzoN&puKPI#g>QZnr2EPrpG<
z^bA0$yA%vN29Wcz#<(US80UEZgy_$&u_AXi8nqf?-H$Mwqw<(-X#^_RVNL9YQW(3j
z65D$M12}nzP~wvdTU_#)4?k|hxv?0GO5^rl`TfL*rGJ<O9ah9HC5?>kO2?(Gc66G#
z3XS!N#=6?mC~dTl^;1cJEHPPHStSBuXPWW-5n~+r?n4ISB1n?)K3w~iJKt|Nrm43C
z+2y^>EPIaN20nMsclEH{J3Og!T{`Zcd>eO~`ck*RYF1pVhu1u^3fD2Ytcu|+2pO*B
zSqH9Xzm`_R5SJ-wtUZXnxdXVj@FxFzS0Jw1<c`alERikc{G<QAu|^x~xLrX)7*?7E
z%NO6`#XOl!TCa7YZ^2~ZTrrjApa_k*Rm66`cforL<w@s&C^dJi!}ODg&Q%vs@9{L~
z&D%jf83FrGd;n&j|H+n03Q$v*o48J!+mmw8oqVoLfn@!u<QH!VTn-X|M!q~vG*V)-
zrV29|S*J14aXx<fv4rZ`OofWr`J`og4S&`wE%evEj(#V{an-O6WS8H?@jniPuO37s
z7xJiQ({Up6-W?LSJun9y#c07p3p#K?7TxV$!KKGH*tp?RWX;8C;H5OQS8yTkJLIW{
z^b$7V!%=qmqv=d!x;NR(J58*!M?pZ>k$?OLA6F?D5`p-K{7$uKhzlQpXA2GJ+&lWz
zMYsYITP_fDlNxZ^sYrkFQ<yc%Sr{A`feu!GxOekyTjLw`@J(X|dP!8m^^W5(>a9f%
z<;0K+nMpj`-ZBt)`0ggU32?bJQ?6OK2NUUb`ok>>r5A}}?^I1f15?T73`;C|+{zTZ
zmByG2SDE(N*KkYkQB1Zu2djCqAnB|}vvlII#K4v(*CGYqa<-uT?+Ea1sKZ&;55Rjp
z6H5Lr<hVK+fMO-O$6bbgs8C_LPaP%wA-eEvzzfQx)ag0c!g|{2(MPe0G+z2U-VoBH
zYklQt*)dIa{rg>{Eb|9{B=!uc{jm|7o!&EEgN9gIt4`HCo-x&`i*aD=9qM0<$0K`B
zfo`fWEiAl)2j6jdwgqNv-r;T7=*iHMjUV}^cbb7u&qXq86QBS2NeMi3H^Zwc5g;|y
z9?$(}K%`Z?$uo`v4(PHaBzZQf-cTe}ew=gG^(nfVCv!WHUh``W(ui(qC4QKegko_%
zIPm5N=FD`bEAPfJ7Tt9yD`-PEEWLoX-Rk(Ueg&2WWML?mgQ|}!hqZsWY~yN*CF$Dq
zb?`?lp6N)5vp5c)K{QsFL0kR{qAhhJ%$pEd=y-G;4MV514Fbk&p=t+~x$DsVdBRLb
zm^=vxU5baTD#1iy4Q5PLAi0mG!6Xww+J0jN<FsKvtC>C*_M09jTK6pQ=8swQ&h|Pc
zs(B|}wEZazo_`H{O-A8W9&&E9Nf7ru5d`iDlDv>7m@j&UOYZ8?aP8^1OuZaGM;el}
zG97NW%^3Qr<iH#6b0i}^j8)f2BBmJ;=yReTR{rs%5%>4mPMW!gO-NgUkzB1<sMZFi
zWydf-*NZryBza&WMH3IHkm{8+(Cx0y7upeo4P3@Yep~_uDj(y6xgTNPTp@ao>v0rb
zHN&H>BE-7vC%@hJ3P{IJVm2P<n1@GtER#M3?OR`=>?7_O@ui<PV(LOw97iEoj=OgB
z-av@Qd0Y4LG^V>}I(157;l$H&REY4VbNhqn%&A3W%x#<rw))1b>#WD2k+q<xl?fTg
z%pl8J5F=l)Saqre-!(^KR`(~Y`J&7#Gdl&!L!a5+zNO5Nfj?7GA<3A&d%;TDSfJz~
zClWDDfZicHz;@FyjI(LSuXCzkrBn%6oksRw=T%hoaKWmSR<`~8G^)9mbD!kK;+m_k
zaLE-}n$dZKX{TIG)Os7ey~Ls9K^|XY-b{j*)XBG-Ghly8Col4R3>Ka#hB-EeapUhB
zjFpEL4fQv|jvLo-c7zC>ZgicTwB@|nKUHw9?rpe(pPAr$O>Dr9h1`Cw^H{0jP9AN3
ziH8)XP_OMWBzaZ`syf_9@00*sT62dL{Tjwf$Cjh=WHlV$y`GpAnbIGJmk?L3M_v#n
zM=z@y5UF_z+_hbcGd30?^?$+&trDSw+2w2&=hjWVq0CfXdW$jBBT2=<HV7?!#E!^~
zuom?KXtwekR((%kg!8T<t9u{kXcY5qKD)(Oc6$-u+w;gI{tC!V<7%Jv3S?cb4J$UY
zlxUw_gHG*YWPRr=*k&yVjj~o`yM-kv$_t=N%T-(y5KVO-|6wQLeRgHFC2emorFZHB
ziRUjxlxodnPam$tE9NOg;7m|;*^&-h`+Iq;h|og3c>5+ZBq&c;Xcpp|!~~SAodU5#
zK`{2x7sZ6%G4u0xbGx8sP{&E;<Xy@iR2;buvrg(TvC&#YM5h3X1J9$IR5JGOzs6;T
zLWusb3fZr9l(CEbfbyft*pO6<FT*RCtOtu}=<6AP_fol?kVA~>L>*Ew(8~GCi`dkP
zYkZ%Z|G;9$ez2Tbz&>7RO>|!C5r>_diP{Zeda*>5#0Bb6qYa}t^`#ViPJIm#+ZAk^
zqXl7hH0M<tyTpvmSA#X#Tqfka9<J6ELD8@JP?0$ZA~hpyi<%3|z9sbKy>wi==q}6+
zRL5q|*YGyxE=;(_?SN}N2gc?DY`Ew$S`(g$dkY+3)fELO-<(MbQ@n|=_Ed;Xk)u2R
zzJUQQW3uz=OD1qjBe$P(5x%@CMne^+qN|Y?ZOUeu^<i0@Z(W(xaJxY)t3t?!;V9y=
z<^%#ymkjh?#Hky`u`9BVRaEvMqrBftj1w>>ns&6cn|t<^%7cqREfafH1$w0f;KZy)
zsC)lC>b7-)e3B!zKO}_8l{`8{IvrpCkiz)>WHxoM1cUZJ#)D?dQPWU{{v9}pqq6;M
z*1@abmr=z(^1K5~ObqA{cWotd`|0fVuVSD4=gR~SyTXcL0KMdsxOKA-nERXuwGEpv
z@A-Pte}59bHGhRMZIaaP822vrALf7mFC2D{EW!%yQkbp!6nFSqlBk1|$;Sg0M7t@2
zB)pYj1kBQS6S#dF4eKUAW1u$mHBh1Zz2<YdyP0IYg$?Cznn=1#gy>%5SC})B%wF)h
zPHY7y(7~LApgF1oKi4V{ansMN&e>i(vFko}j$RFOFPMOXC&w||cE#zH24tJ)b|{F7
zXKcz;X;riWjoK5>_<a3{f==9yn%XJEaPM>s)t-XuO2WA;ImK^PlOX+iEEs24lg<;@
z;aL|q8@#fR?iA0)IMXVS=ih|U2vJPzdc!mSI+p})OT_Lb&Y8c$0^&KA`(Vxza&wsw
z9h<U{9_Jl~%btSt`1(xHar**qQ){uQaRXGGu3#5AZX()0Es3M&Z`5kBgN5HcXoFxA
ztJqiu1!wo-mbys1aZnx%6dvQH7!5k>7&k{e(u=jB>TFGn8&RBp0#<c(G7Vl$_)udv
zM)W1K%gH(9U*1Nm%x>VQuMWMi<{3KY$D;8<Oa9N{d#vPEP1LVjg?3htVekA@82I}O
zkF1oVl3(kYZ(m=bV`~a~|NS8leJ9GXyKQ`r3+lAo#TK2FmeXJL3UE6)n)nyqM3JN=
z^xiZT9GaWJss@GPEnXIWlK~ufn!!*}6G*z$hP!fGc@f!p{BOq>P`yKQQL@;J^G}xo
zGqajCw7Y<Ru3co6@7Xh@1(8_$+m#eQQN@F~st~@T7#sgirK1a6A+bW3+f#EDPa4*M
zyF@#7RC7Ly`L5s+@Q)GuHxV5SUZMt_j;cl9(Qkwy3w{Wa2b=yfmjxB+x*H$af7_>l
zRLwAy{<*?b2Bl)v#d3Dd8g(#!<b{*8Pot>9cJP+#2FnZ~(myi^W0Jag#+6?gy)vNU
zMI)?SfIm5>%lT3wEb)O<BRWO=W#?;0v2Om=T>em+Q`TNbOCN8l|BGV%DjxkjSBsq6
zK7`44s-RfI0ZSHtXFVci>5m0VP(M_eu1&iLW8zcjq>Oo_-fAy1>5CZZeq9Kjw}hBy
zt`fvk&WieYb1Z_;3WzqC%JdmXU_$yWUTR)GY;`P#PuC+zl}Zt=Fqw<8FQn0HUJnii
zbumpZ%P{BfW>PsUMOY6-VrH3-2?aSs`A-U-O4`afwO&Eop7VI912C(mpAij`!=pM2
zFd;&P)ST_*+xy1i^f>}#;Qbryx-UmP^q$~%=K-d>wwPxmrvSZ+dvK_D0N1H|py5m&
zVLug<^q%XGtRBUT?>Gy0xc69KhY<YW-j6$XaWxJj&dgtFN=x<^<3Yg`qPXfl=v`CA
zQO=XH$Wj*H%U@^h2kMz}j#~*lD}i4W>haR0eDoBPCeCaBq06W{s3m%{20`y2Syh!9
zJ3PiuMtrs+T8r^;<Z2QtWnyTVKpf{KqEJK|8ycR;)YeJh#HOuqwNsXkHY}l1=XB{L
z;Vx#;W*>a{^bOzdx-g?VXbS?blNeitQ=qDz3zZcQ@KoC)CZ2tZ+6pZgOuyiw=Tda`
zvLk3W-~l3H>df$MWwNCs7dQMTKv$^6VN=swI+bIhw=dX2{T3(L$~=xEF8%LOCBTB_
zk9ZS%<w`u6y^fa6wc{0<U%@zu1!xl*&31b$(4EgCQB%*6kj2;FHOC_#Ok*%Pmg5C3
z#$aWr3c95&r5WB5<Xg~F4E8snBh!UwimEr=^zjVQny84wa%WI&eLr}bOr*+}cCx`o
zB}w`BUF`n57a;EDZRQHMKgaYpi_WLyFu$l8T4ogxyG7ec>?=jO;$s#WTVP3Nb&7Gb
zru+EdEY}Fu6D1<0RhT)y6O-zH0KdNzv>I8s;W3DMuhV(sD-7{jojEaZPC(~*!#J@0
z2nw7nveD5{Ab0-A;?yh+dMrZ%Z~qwr+m~8I(oGXWR8#qG(~Pmp`vVp+Y2@%17Wn%g
zK}fY0RVg>6@*#ilm)u|GOY;|~DoI7-98FRa^9TaI2~nf^Q7%_o#(WO5gvDL8@Vfdo
z*Y}-I;$#y!{;CuwIG01N|0x)~pg{ww!-$Tj5a~FsOjc+=M&;)NY=XcN(y?|8>L;Ay
z(~a5OoGk&H^j2cMsXoavK8N+!U$b_JeuQ;(gw@W{v^n`bJ8hFHyS&Jdk#P~Co0m|w
z*P<PTejZ`lcV%LC@N%B@Np8>4zWW&Aqrn`$vz>{l*P(9L)zEe#A7|t&LCtG(AhGuy
z$aaa5u&H_EkF61E6t&?8<`gq1dYp3v*y8W$<FGZX4e5W9#Cf6$24AU%O)VP4E7*f}
z`CLK>X#k}v7jlN{@yER<XczFA7w>ltE^{oG$(~#6_-+w8@Mta@79&do>+i7-n<CLH
ze+)%!tVn280krO(PNO;hyifO1wys-&tgqa}x&`@=D&1rFrX&ef=gWdXg%!C{|CBxT
zEeZe3jY6Y96AZZQLG~sTLusTd4fw@*(^`2Pe>}$83z%d4^}S^JMPFiFIgR?oKV^S4
ziBmz2+mG>;qyi63$-P(YXncX&d8ZS@Hkzd2se{+yG1uEl3OmZ)h~l!+BnvOT3?o82
zb&2iBF;=a!8e65BFpt&Z)?OtM^Zid5=b7qM=ui*|xl+p*iwL2$PYu*qh1=#|n@K~3
z^Z44oe_;2c9@~NSDX=KkiiWJG7@Hr5<a8$+<(UIDuVsl2=h5pqp+HLy#-eM3ELA!>
z5%(!`JrSpB)>(K0)8qIa?>Mx;y4lN!^W;M)7Nkp#$jm^I?^;xG)oMo0RGn(^=8=Zd
zVJP%dq^Xm?<I9po?Aa!7ymm<vnAmR^+Y^b>%PAyE=Hc!0dJuc?HuBFOCR^$^vg-|P
zAUB8eeH_VxBaZXvte>K6Lh2y1_sB-p&(4+qXl^myP>h8D5nWgq8cf6s4zm}gOrZHY
zxE$uMqkLg~hCZ+spnaz+a8dJIo?D9w+5{?)dv70N%lB%$cWNd+d&X_3=6>H`I*DYa
z*I~u%e)dbxSLELprE5O^WGjCJ(rp1k<X~1BR~sB;%yhCL*}<3|QB%iD;l}i|?l}Be
zsEvuoGns8#>LhYrDD1QR4GA8B*w)mIo?A3Y{?t<ZzBLo&t(MVkPTW2EegX<y{(aN3
zl%PqM107x~OiJT|K>V>6Z0Ns_p6j{&T8+wNb|{yb;~4Ga!Ei`lrULbI`Y}sv0!1$u
zB4aLw)|{^?t*RN=XjjyGJBdhfj8%GA2xeP<W170<!BnLTluBP9IkOW(7Tdxe{az56
z@{|2ge2O``;xmpO$O6ObA*^;g5A2#B+_d`n7vs*}!15nyK(|J)yKG{Z#B@_uaD^n*
zpAbdLuU%rj5*W08BS`aJ{O0_Yn)LoHd-`mOC>cF!P6J8=sa~`(DHZ;VPF|B~^ocBX
z^~pN8Q?r_Em6<`EN{<tB)k#G8%xixB0%Njzp*xr1zKac3Dj1N($&mV+@v=Z2WZ2*3
z|Ftv%+YfJ8ZErbhf5VC1$wW5ib|GjN?!YO}8<4Z%;)eo01Ufk($X|k&p*0v}d55*X
zlEp?$PqV$SOMtb^e2N-g60GCgr6d%$5oNzkw0GBe%<$BK8T<6<fh%`eH;(%ZVz1)k
zjn~+>Z#>DjL(&X$^%e^H%m(K*F*xh}7v7EMab#0wBI&i@v$od?Fm6nld<{a#Us?t_
zHE*jM>IC4*5@n*iCYh1bGa$eE|6<Vt6?BxyW2Buk*n`6nXx`;Q5_LzQ_Qer=X0-#J
zzt_gSe|!+Q{%LK83;p!19`kk><G-SAj8wXUv(Ae!y)!gmouEEdnqPzK4y2Llr7Ou%
z>v70D-h{u$S7C|6YR1YISf{LiIB#e@DLdMLP}h!cxt&Jym5Q-yf&l&;<DN|}i#dm`
z7$ZKv4{v-F$G124ke=zf@M-TFNIxIXTqt13%|8M(_~Z#@{61Hc=<a}xd&A(^eQOfE
z_9yds*E76%w-Qdi<2c~SGiZ+9YBpzgC^pRFvlA?S!eOrzF5kMAev!V9ZqF98BVRAT
z_t{Hm^QC0SYHPypp_DNkF`<{9y~3}$A7FOgM5>tj5=Y`wuqo^<484rzIH)rG?yt%C
zZ`C8I)Cd&9-?EG3y3r{7A>J|@g01&gzyQ}bofYKDE($Av`5f!Ebc-4NZ}&M2Ns0w?
zUBFEXzhTAnAq)&Z0Xi9RypaY$QnmCxipYNfiMrqTWwtZ3lFOdEJ+R~QWyNgKjTCeq
z=6L6nlMqv!idKzVh})c4qImBgxQuc4*M#GwQ0fY}uLy>i|50@2Q8|8H7;c_RlS)#l
zq>@U~^gjCtNg`5aNrqoY2q6iTB#F{YNs=TXNz!}vDTxv?CxleykW3-@&iAj?TC%+F
z^PIEyeP5SpG|$u|(cnKXVeF?ieEj>IC_LGZ-gb?{J6;K5$=OHX?Z+T8NL+;ZKWy2$
za~^1{8%JNQR%HkC;>d3=ZM=Bi76;g#B!_s8CHK5KIb(5~mT?Bz;dxou^DTqOiPLD_
zUUkgfq|aJ{xubO90E}tPhg=0Qb?CSZHOVVMWt^=TuF2zYo^i^`$`|&1nhSa+=U|2H
zYhk_IIh^vx9$bekf&FqF80vUlG)=<-O9y@xX)EyV#xhg#w(kY<h$|N6_i!&r7JiN8
zY@?tR7}ef_ijL1A&Qc(83(RnD=|XS|Ucg>BHDRB<F=(%qhuUc?S?HKz@wJ@KICYsA
z=LHR7xn^!m2zrYd-*3S8Sni|qJc0?rOtL=4n%#`(#x{)sZ1Zz>)O%ZqHo3JJIqE&{
z1kV-pJa5zVnfJk^;y2uC@?iP{Yfz(H9<Q%mjBY-)=>E4zXl%WR$%7qe&#Eq%+AWC_
z#u*cd+EJo-b3HbcYQf<h8}Qh{6KG09K_N6u*zz-zus;uB%j37`6_!H7KU;AB_BVXF
zG#w3MeqdR^88YEzG+A`b5L?Gx!KM>B%>0}#eAA8M&S1_CmKjWRhBK<ac@ZR+x<l--
zxx&uYXf&Gp9;daXk|CT+ZczMzh87HFImIGaQD{cSSNd@$eT*pNhdl}Ae^>r%&WD<0
zO<dKTg{r&Zn6zGrd0Z(YM%|L+tb-X#_?84h<1nJed2E)Evh49P4~UmI0@49va7pn~
zD&dd?&m2vOT~IrQ-z*pUjgCO)^Z(%LK3O6u6M%hTW6<;HR$6?`37o@UVEm1VEM|QG
z=S+SPe~=hRO!j4<kG(d};Ps=2-U^aEULG#GKEc8-71)3e#ft*#1*viu@_1x19NTXx
zRJfdlsHp#hIXOHpF!w)<UE{z`NEeGjr)<PWC4b=S{+rmH^_*)ZCt?b7#q{O7A%4V1
zboYHp9Yt$cN9H-Wp+1~#+mMT&f&wt<?F2}$%op_TNwJp4CGb?-3&{hn@!9EjVa2nF
z<Z-JU6E8O-%IpKSw%ZenM+w+0Kbd3=_ylXFTaaA;K(gkyG0Tq~Lta)G^X;6#WX?LG
z|B#b(I7!6S_@5AFo5X)U8`d2Ug>-t6)`brjl&s@O|EyR%weT>Mx$i~K>ICptUQDJL
zNU_D7<(+776<n6a(}#U(?2zVN;dX;BR8DK5$6^v;TGCEB@%$hx`nVrbb$R~#^DT6U
zyd>%>)q`Zy)AXIA2_&Z+5teUyfQJqGaL6_jcFx*~J>7Da-a9^(_~=WLCBL8H?!RM5
z)PYpjQiG+!yWO^^)SQAL!ME__@km(4&m55ntLd;OuQ(vpmL+R$$NlcP!iIQR;%ya=
zm-@AsY5xa&GWjE|v8e$3Z!9UltcJ1@`ytBcg6P8>!rmL&lj-*3$Oq2+dRy>LTw<~r
z#OGJxT+bP#I3x{|D~*IZ64&tSPft?CqWG+|iT1ekqrd(+&MIDr@7_+tcC{gp^zIFG
zCud`v`4{ZXd?*e!35I=Lx~%l31oPXHK@Agh$;{<qTC3Cnr%Q(8_AB$5*C2H!Gfj@w
zkDEg3&_#&VAJ1noZK!U~xeLXG^xs5h2&%FmbON6ZH*yv*?{f`&yi8a-+LZ(~Si+*#
zFlM$e13u0Ff!bZUf-2|8PKrH+Ee_eZ_J}|CDi*?qOWLe%&LCW>x}A)?@BvDHzZOM_
zR=M8i?2sc<M>97iU3O|FLd~Irc(Cq6)j!i07;Zh2de+KuUKwFDtp`F?V%W56jkq>g
z3q3;ip?<Fki7qcdnO*s~B4a<s{!}N$X}tSuasULY5oFc0yEOW34b9EHh9gtVQNObs
zI*oML0l#S&I)t!GoE7<YrZMr)v}8#atKdyhI2|VIOyf0Nv2FV*lzesxJ|^ap(T|6-
z6Sc>o<7z7UZd*qrHNvpMxe1b%O0%vSRkC#y!F3DO$a51*zE2rJP94q`c|M&ibWW*7
zmF-(dguWLE^xQARo6JXAYrz7K_d<TMI&&MpN!VoZnA-HI3(A%Uu(RO{JYSYXswX?M
z88XVG^=KR9<bDL%=n%5~iY=oV+U(b!(=h1mOFW|*O{NWwK{K-yRQbOfsA`<x!*}|j
zcDFF^)G#!9Ig;+1mIHH+d>8lnPhveO@kj@uuu|MfM<{9F=#@bvVcH2$$$JS7ixxuc
z#?9j7bScsoH$u2JdmV9ZpGp!#)8I?Z0<2!|Mq)0?ptGJTD>pYqXPa(xcDz8Rx<$gb
z@lWujtS;yA)?n^mX|}yxmz*4-0Lo8BVDLX_@;Tm`#hWOA@Xr~FN+hU`Ts=J!?|^3)
zxnR1DBpR!H6SMC-X|2s{?8vZ$l4Uz_((jQ(cG+l>CaFhMp6Vh~m_=tyAIRKRJ;%|W
z!!fPHghg$=2~k6IUF8N@kk27s@vp{HELdsGmb?nW8Bsa#$>0R~PJ94C+1tP_Bux-y
z$dcz_|LB81B``)}8qtWi1;<t`rub_t+dk5aJbs!<^zSx-+IlOLYB?m_T4;bWwYH??
z&N$NEe+_e*vO(p<db+ve3i{Ue;>4-e&}K_m!<z;grCpDaSGtikt%lnZ>M(Z17o7d2
zgZA*elx4<y(8|6`k1Tm9#2?x%E--e4&A-i2ZKsH-a*kfrA6ZDdE=k(V#<K=pb95i<
zO=eu0%;p@LOfn`IU|?w*vJyvjr#=(4gAo<3&SX(j%-FrMv8ey3fJ|B!1w$87)c&!O
z4!vT?{f%>|KD~?|eN*7dCtVg6wI8I-ZJGb^$EbhW0$X1?Fq79GK;o1$Gk)Sn6?8Jd
z-E5Vxp=CayS#~7Y{yGhwA%j{nomgRT8pYEncCDUCTu(RiUH1u$XSw1Bx5wZYO&MlA
z%N=!Mn!xR~BJ_-rWTBA(Bx|Y~+|D1)bJlM_P4_BQxvkD_zjnq=P0DPptp;iNe}1p9
zh(@WWi8Y;eK=1QdY?$pzJb56Z==u>n<T#(qX;5I&#vb?|EFty<lC0b>6DQ5O5B~S&
zl8)i>?7HL#Cgnd4*4!J)jIWvy&x)Iv{NMojxp5*Icik3#bytG1{dRi&{Q<G_gG2ab
z{Cb%6tr!Ew6C%~OgH6?o#+o<U%s3>9<ZV?F8T`5-2otx!cgG=oZpdiAWFL4Jo08V{
zLg8lgI(QItkH+kpiF@bpuAQqrbF@eT_5Ek@_@~i?ycoiqIA31QLtuJqcA{~do5*eX
zeBMnHW6tP(!rRPZI`gN9bXTN`rF<9Aoy~@1Re31PubU4IW0lA;V@0<8Wj1vPsus@j
z>_|qC98UR{ME+dy5L*wHN3WZMNZ!4l;M`)!YMj@Kp1q6#C*5MmDtae!edtR<12zk5
zd1fQtDhFiG*U>hsd*a0!Be3wlN}N@^hTjwASXtsL2>i|Pj`>IU@Mkn}SB(?pACSiK
zYGt@H{3cy<p8G~$#$cntNf<Q0mn!A|LzRmosDagU!8)yi-n@`Ye&%%Jon^s7ZAdKW
zTaL${r;B0csd23Dr7Fo^CXcHwx)QI729bJNqv&(4F6O}ulGhN6%ePygY5#T9Js^)&
zff~?j?N1H1y{1+rW@Jh$pM^%7QthWn_}*2I4Os9OLg{j%o|aB$H;-jL1v|;vlOl4p
z@gV-)y9ct@B=GxuCHD)ai1TFg#D(6?=riiGXs2c>O<x0S`yM|W_xCL<$heHRo77pm
zdLihk_hFZ49VQ(4j=d7w$;}VV{G#Vc-|X}PgS@eVg~?oj*gxi;^{Hh1!&j)?@J$H+
z_!;ZF+ThFwZ-UY)7#ix#LaMh4QwEKoyMAX7_0y3&&lNAa70GKsB`;vrn?Bk*HC-4V
zzl%<uJ&Yu5xj~BumtaqrG5$2sWXBVEKlR5+I>J^F6zo+5iLi;6TqKsUmamhE!R}Oy
z+|Vu9RlO2g6_qj6i2uK(k7K1PK8wO1_~F57FGjW1sl?dms?Qf>NsrN2VXW$9G(N1u
zB<(&zV_7tL8qSbNZie+gYq2P1B$iLIhaKjbuw>yolrXFit?Af}JaB_GFOOrB_C<U-
zY%{#K8I4~9@?idjF6y4)3XbEClcqKWFgA;!{hMsyjQB2I=e?H)9ml9wOE~<S@E>tC
zTSFGrxq+UZI@HNNz`koO_#@~9_Q+<#TSrM||1Ff(IPG#(ePd19&o|MTgJuffqA$=B
z{!t`NT*9-F^6YOUKY!Tiu^Ap6aC1u^sByNT_;DyvTv12A&6@>dW4uYtXGM|oB^&yy
z+#Baljl`}`O*pd6nzRJ25F{O~iG-J=FmjzOO68pu95>lPcjzIp#`C+l$eefmQ+lwy
zeL5R^U>Sx!{6+Vi(ZKipj)MQ9HSjyU3e7i8Vq}{Qd-i@fd3!Y%L%R0T`<q9Qw>uMw
zqs)0Sw`39xx3pn)PKU%|?HOd4Y!cqNEKeH6F?4?RJ|fS1oFy?k(OylG+}o9l4NFR}
zaZDVUc4Z(aS;^s#MbA)gu@-YrT!<})^_fJ<L}B|+N9Z>AB;Hr;NV9(oWZizRuyg-B
z@SM9J9)_Pm<z32{cIyoc|Na1eui~C|yHe4?lg8{z;U!cEUrB@4hU3bQ@{IY0fWow|
z@Zq%qZs_IDt|dOg-?BV(GCNDNHLnQEoF>4CPZ=aG%ogpGK0>weQPegGgtoL%Ffec@
zNC(xR<ZCCEr|U=q=b5q4)w4<4!UKHIah|qm?!_XnJGh%aBP*?)nd&Niyy$C&VKGl(
zHNW%BvwM#<i#I`J*G#%Tz!F<Ib0sF^9<i+4i@lr12(3;%Vygo(Son7Y@qDUB9{f^c
zF^a{cL*JGdoz^D>Q??238uCE?;dHh=*My904G})_9i_gHK6!jlo^-l5!87&4#6+u_
zL`jYn<{j~ZSl*XVf6Q6H&ySKlO|Ni{ksn%B%p;?nFNkALjK?Z6fl0hM;?nfpgc$#{
z6bB4U5W*EKVdyFq=6<&WMf2}q*5gNF+hNZz>dQoWy7DZ1fA~UBG2MWzW9x*V!f<L>
zD~q$g+d=54>C|_6D$O>J6`ISh3%j0Zk?lt=qwMQP&>f!&J|9MtfFFr;eA8Dt=e!*|
z(&S2fD|MiKY$0~P2_a$AB4|vmF<7+o4tD%#s9GFDj>iF8<jwtlat~;y;tmq)mL!y%
z(qpX`wi0OxeG=;V4UTs>GtZMFY3ZgO(Z=Zm5Vi^|bi4+WsPA%)B9{fn|3vuOF_Jh{
zR-?Pda$+q`gjh{IG|0Jtc^7NNn(K;T&G6yu_W&mBQ+H&=QkG2rS~^ZVLP_?GW8j`;
z01+xraQ!xIyf#7?Gp$B}YJUrE;_r)|?FhG4PoukPtuX%%kk8NV<F(x97^q>udrBuD
zAomb0h;YPHMkai3PQho=H?hVd7oy;q2t_~2g|0vSc;8l^0pAC;=~jtNbDHqgApQ=r
zJS@8XTNxt)#$n{ubiwUb63M==M?&q*L@Au9@4YvTF4|H?O~(HR^QZ(VOUZ&Y!EdRZ
z-%#+BPi6`S%Ww>z?|6+mjBbBw#MW{JxI1z*Isa!5*%rAMwUbpi3vMp#j@=ER$JGR@
zqb4LiS%r0;a>ru38SKpRbPQ^fWf52Y3L#te3I>x-;g5o$49?pyB~LM#-QY)7{9Q>b
z%C%Wb<~%6Va>AQ04r6`273ak4LMa<bVa%&o2)xfkvzPpX5^q%~{%<b$nlBN*?NVa{
zS1PlA3f%eLNYOcU11wk&M<%C5U_-4E>-eJ%1qoO1K+6QYefv7L(II5&whP#$sZJ7C
zO9^KV_X-Oxa1L;9C;r(n5{T(TfciG*w~PasDQ~#DDunw%rZYwLqok2=-c!0Qkz#vb
zagYg{N;KFq??Jfv;&h&69!-V>EQEU6jE)KR#8LGeM#zj~y+Jx;?(HFD%)DNV8=4G3
z)<)cUVnJGg@94=t!S`c3L<mYu@w5SK{<mLPG4d*g`27dlcK1P_?s-9B&mLhrmZK&2
zy~y0%DH<B$%%&UshMYEKA*@y#l$RA^gUNo}MxOGl?JMwo!C4~_-2ECol*wo*F`q~0
zsl(<AWJ9?*3ohM^jk~x@33Zsd8TW%&R^cq8A;fyGCUc%Wo0SY+jG=BGWc=52P;uM{
zNSZED^ALo(#07kAJrp$VOu(XcEwCutiPr5gByGT8u;SfzB^7mazV}c#lhT6EpS;7!
zjx5YLYzFU&!y&@W78GueW-b2!>^JH`n9f}+I%h&0OCF<_lr4E>b`#%m4&>wmZTMS3
zft9Yw!jeZTNRNpwM*gzI%8xye_FkWSn|y$LnKuBG3ar?jMs*U~>dKbQT0u`t*^MEc
zwpcnb2s~RHXxq{NS~hqP`!Q<`m~Bjko;9bza*ra@CchxMd@rn@(T(;-XYj0LBDyb<
z#!Wv{iC&ij%$NT~jhfYj2kR2Z_%Q>po{qqvBc{ZRGZ8BnYJuft1!i79j;*+^j90`*
zp~B}jNT+TirHnhzDys2Ee;57RtWI<lnlPhM8f!ir6%zMOW#a-Rn6^v>>~WVP^{cH>
zUE5aZ-S!$fTbJ;;Lb1rjz?p^K-Y0I}R*YZ%49BXcR?txPfgWGJo<P+TynYH;)b``9
zqx8a1rJXbAR(GIBRkz^ToCm-96^YLC>%#V(pCM&cChA<6igs>u#1`3^qKnHlSWGtJ
z%sK~Lkz<X%q3cksSe_M}nE??Q{5ju#3{_Ty(f6TQ<kn6*mhHZis5}e7N{QQ?V|*K$
z1AVdTCg)+FH)P8-t=PT7<>ZZpKeNiQCKW@6fx%i^hz<zD=Bz8+uau6}$vL>SAr7_7
z_MnD)Gj{g;zz<4a;9loXn$l{9_N9BN&Ath+S<xJ8gUp!P)W0x!*K46*e<5xR<(|ty
zSD_*?0|T$hvn7kq!Jn}|P(jUx=q=|inVK0iGES4J{^otHV|!qf(Mf7Q!I-?46Qkml
z!Aw5bl=<wMhH~E>P@?xO%-)y<iahHVJ@*i-coa>t(|uSQ7>LVG%Mv*$2~_PLN4g?p
zSlLBjCI<@0hH(SQ@3)~OiAm$*0pr*jZ@wQX8pLMJX#yM7U+`Y}FkY0rh%-LOu<(_J
z#6>p__XWMcZ<lTgn%=hfGV2A-2#<qzqvnEITpoUq+)UJS3aNJbU^-uF2nnv0A!i#5
zS&YOYk|jM9&yC|V;Q(*u7~{@VRz1ZJCpDOHfSjOpUX6%WZYLt{QnNDJ35NA$f*t2C
zC!bLhXSv-MKlx!pOy=`hYSMW=ug$_;t2d)_W)EJjAH$X{E)nwW4}wX4n-KE*q@X$`
zi&pnru&NECFmY`P(JP7<H{{sxJ3a5_y}E*-gDc^2=yQ?GKu;+0>u}ZTn9PQpO+<xT
z7xBfhA{;t%FiQTZ!{<vGDh`(<M}8&19`0cc5wAhF3Tg7-1!p0|^#B<>hRtXzCC&p9
zvCelVSvbs!I^WY{nFe#<8q*=&><q?OZRXy*4eaLqeH2=Z$gzlibU&lQjH{Jc<T37?
zxm1rP6`b#Q<QX10HH+@_e~4wXBO!L|K<FP`4<q)D#Ke7%p`5$<bYwjQwIM&T|NTSw
zwMd1%);GbF>ICj?%f#`IZ(#H46kPCF87&VPvXd(#z~t|L=(_$49XLLkcI4<V71t>w
zC*6Q~?@GejwJ%V5ege2TX^`OQCfFsl3}=n?L$6!6XrKQs!BN^1Ozn7AuY4~JSf)p|
z#b}YLA<wXiU53|p!XfiyJN_P^k1DRJL^gOR=~NVmm(nm&@MjEJr}2Wbf_KByQW@UK
z7zf*(+R!!0n)}h1u;9Q4Pz+?aPj?4y>og`|S2fTpX9ydh-6o9IKMi+MeIcING2hRL
zxXET0Zj!wNF8eiL)QU0`56;8{uPE|j<{GT3H79Y4SFs5^vu5ky!V*W7fy~BuI&R%d
zbWG%YD}FZ*_?Js#<oeL-+Emof{z*f<d$H_5D4zejjdYpBQJIs{EVN^pV63u@`qt~Q
zsN!K%^MnSA<bAA7l|Gm-`4VjOkZ0=>AAtHWC8D-0o^*OP;@=-?#3~?`yf8LqdLPwT
z_w*9Tu#h4z?n$y4YR>4mIi3b>ZpK;3%g8jH5yCMVf$?FzXb~>M?rlh=#=rX@OuHG&
zwf*39VGVkEAEIv^O|jCI;<ooH%mS(exmnzENN!+%j6JCl){1pU7NT^V3`<NmfXGr=
zCjBKGZNDqPj-DJGqnUtpqps3~>I`yw;0Ii`*oL(|@g(k%b=V&k2mb0OX!{*Qj8Q!W
zt1A<+b+bN^?Cb|>v<tU;OQYJJ0Bl|p1y?wKWUTBZO#5Jjzdv=LOMwSoR?EU;M~X;p
zj1zfsb{tc8y(dnSHl|0c_X@vGO5?J%(&YB>ecU~+jdrco;Jwid?Nlphcm7aj%HLOU
zhF68^8}G36odRnyu%S`%J6unzrvoPn!-Qxn@-%QH^Vq)u=Se2R-NUljHL?^Pl+wjb
zva-<FY624f`8w060{B*V89Wz_NBs8;!ak|8Pm8%H(l;3&IiJAs{hNs3?ErIJX0uZ2
zD~@Y2BTP$#5_45V%NwE~OJ)~d^Vo}pmkwgmy)+DKvL%-8d=K;NJ?Q;&WSUN#8RjvY
z^$zS4k^DJmG5k3kJ<0#Q=Cj!BrFU?l)OnCN8sl0#;23Ut@&>m}`iy1w8nC%Nn4|<s
zv7KM0;IKcvF#pS6T+;dk=HIHIvy!-n#EgPM?Kw1B&_Jj0-gR-i8FOn&6<zz7fdlo7
zP<fIP^EJ7PM~6n?)9f2Kt+)@)Twa9Xe+SSso_|YIlVE!;PJ{rNLs;5jNtVc|VgLGD
zBxqzV2IVaVWA9MxtWy!>x9T&8q&i%<O`6?z=e%Egb#n0GOqi}ziBsd{$Y#zyAN^?{
zDgBfzIx97t?>q()wXPD#|8oPpf-NxaYcX0^%z>cf5R7;=hpZSsj06>R<5x))w)Fl)
zkasi1z=TxlKKu}N`5dD@;%-_nzK(kW$K%elIJ~LlPp9n=z;ka3&U5sE`D`W4s{aa(
z%1%TfFA6nymZ5prWETHnji{Y%Br0R}!mi#H(A{kdMK7LG-8y}kp*agC*S-fwp6xOZ
zx+^}jTmkaO+=X7Xcyt?O3il3<K}ppKJUey|*;F5#GK9OwhHMcI1S{k3_kP0rPDS=C
z=NsJ86-Y1nA-YqnCj|c2E0Xd#j7|l%j7@2$E)&nes<6Gp^QtOazv&Y966ld@{R5y~
ztr8^+UI`l|d46NrP_}YP6?m#hkm;P~BR5EvRiqC?wI8pbWUUsOC2Lb}A%(c?zKqYp
zzT%h#7WC|lh2Rpe$aI&-!KFkKrePR@lG2(;uRe$3`-kw%^FH)98%|4>MPr_ZhpWY$
zB9W6k_onGy!&i|W=*oX4r(PdjcNmj0zt3=@JPlG_yI|7Ke9TGFV4e>eMKnc*O^xF&
zle70ReAHv{owqN<ixo>S($k#ej(;e28yHO%FYUr|eO;z?bu9lKD5lPng4{ViSUTY}
zep8JhGfrh;;En*{=-oe9*B600tK^v`?~D!dnT0nur-5hnTl(zlS1jmyE*NAglJyas
zHCb{4wzwvPRrplQUnRpjXTG5i21^m`xG69qM2&@QO@y~8IimLk#w=dfg<TJf#>2ZA
zMr5AA*B(8faUzhMkbDN?KHUV>buq%?pSS4vNh|1TiAH?Cpoz*j#uJD+iAhoV`1RBP
zjC?VT#8qpv`V}wGJhh*`*RBS+yB@@SfGRm$Y)1AeFTlO&$H9Gx3~Q@?fc`h1LiLjx
zc<tfBeoE|vG=Z?MOC`v=VN0-@dq;b>e#A3AgW!ZrF7z$=hdMI_RQr~N?|1LzJ)mqD
z9y^lgY~DbAtA3|N3$|jhLyGuG`Wf`EoDL&*@57_Jw~+~)Nou)Kir*)7+5C&y7&){B
z^E0QQ`}%0?GYAIblnb=;kde@I%MF`{a!0Mm8#K=y6q@~YnESJbaFJExv?Eb?1WusE
z=}-)tGLY3e83=k?YOwCVJct|0zmtx$nDj~$n0c}a8~u67kMD5JG&*2Yq!#(l+m!5-
zF9!GDw}cTU*TIQbG(T#Y;ovnJ$sO?`3^j9P0X63_cvT5L?B0W?f5{QM^#_DlK8vSo
zT50afwSpGsZ#%}tlbWt{Y@fZCdd`0?{MOCD;1xC0F*1UD8|(x+Rp)R(eIQnyX+|gB
z$9>_v2lwoV#-b=QVzZ62KYr9glh*`1G&vAAo7k}e|HVSU#IaELunQC}n=qFYRnRY%
zV5QE%teOA0f2kVH{oX8uSr?#jL^9485Cp&8r@|}AjbvN>54=W=aI@`SICf5%NS>6(
zm45wbWvos-_oqUtv>&l+ybQTRzY3xIB*;KdKYS-h<CH}&@bi-w++8^d7nobH+7xs4
z;xNy%S;V+H-%Q531*gy=WxTLyl`9!hQx5Cp`>|&J2T}Okk?bHR$W1oRM2iU&HS)7a
zrA(rzicj|2c*gE<=?u`}JBGM5BUt-JTh_cxngz!mK=7VOD$k50uHjkq;GlzW$E8?g
zv0Fk2-anpnAKOYobF^9d{TQ5)ya!!>HQ}e1=V;~D<Dxr?Vd9vx+aTQI1r|Jw!!EO9
zWWFzVu#)=}y~|N$L8;(ylHvRZ&w(_!kV@6nFr*>|{`|9K$pKXuZEa8L?KD{3wnHdA
zXAvkIC`7u^pT%8^7mB?8f|}nud^4{LEUJFvZPO0i&$EG}c_&ZHDhJI>lXwPrC0K{u
zgz$n?q2(yWOPnFSd)Pk=_`~1tN#|gFdmC1*tw+D*htO}sTCkmb2Xn{v<C2zYp>1TX
ztJlCqBx=`c*BOtm@jmYy?EIuETC(mneLG8r+;17p0t02l>t5Z&MwdaHmu<|FkL)5<
zoF`ZNaRB)`{2c1Puj6}38H`wdgPISpW|>;9g55<&cGgfIE`I!hnF#?f{Ys_~uu+Yb
zPP_wK?eozy{IXa(U_btfPbQ%@oA7#wC3(`SBrLt;iJKygh}MN6Y`tLsooOmb?sH#E
z1a~=2Ny!&>3=hBshsU7cIGD7!uE3r{QmnDJRIui~H5>X2vbKE|bjOrpzu8J~)*8u_
zKP_VQ8@@oO>?E?w(S<SZdMr39pv#!StlP*JKThPG%3msEu8Tc&9`{*1uDXOiGk0VP
zi>5$qq(Gw%op5a!@<ufDm^w)^O2y}6B|tr{9sNhUk-8A>oZqz&&P`Qeze6}LHc|)c
z`SVM^KN0_~PsKNeoPq!J5NXL8#e^9i;uBXN2n9NI!u;uzh}-RMEQ~b~+PCk*K(<=k
z{=$h3(XGPmx7ApJHs|6_oB;z$MqpZZKXqQR0|!bP5xFyO@n$Xs^~VW9^`169OBX=K
z;5U+mOIWDp4M8$g31b?jGlj+oAxX!Xc&%1u*A>4ak&nX7uP;(j?rU68ze+IJ&pTib
zlxdyo8X9^~Qe^n584pG5u2LIh21HJsEo_X05dZ1qP2>PF-QhaQseTf=Uk(;zx5uIS
z{-daV|0~RQs6@@iej&|$r=Ys|6t?nxm)SF4_GpD8lRrLyU3OPw1t(RoXzE~?#vfw8
z$G*W|qLCoMo?MEFa3Eki9dE`tVh_)=#w&1UNx&3Rlhg#uZOW<rBPqP+tB3C=q{3=v
z13YT`0On3SB-VLh!ruH#qm%zzj%$j3pjD3q3%DK!Z~V8yglC-R>%JQU3LoOr^*f0J
zZ&s>5_()Asi$Fs|m3R-HN_rH{#HSOZF)Z^cdfs)wtVC1xK=%obn9%}#)>gtp(_Q%E
zhXnq)=u709E!ehaS#;dpY^cgoCTA<th=u<b>UCRzWu(4Ez3VOX*|s7o>PkQ<D-YP{
z7!NNN8@SFnau?rU_$NedDHjSdG}(^<U*J#s60jRtfjata!h_l<e6ZmFdh}+HhI(_#
z>&8&#a0+{TEXmAnH@LZIBrAU#AgFs>6UQ%)rjN}x;NvC7anTXvtke}`UMTM;)--|P
z??t3WqwA9H?^aw^%J=DPFWwxTLCPhcfZVTVcvz;0rd*nXDhKtL&9f++_v-*=6%HY%
z@{QPym6_z${S5rZGq}6<m7wabZJ6~(nMDrR4f{u(q4L3s?B|+(sLKDz`<G5^&6#1W
zUP_<bIc^HYzq3izj&zb`_C;u{?!)zG_d)1|(d_GV&Hx%;hD@Z62bYGiNfXO4e9lU;
z{#S%B|4lpwn!Lcj6a6ux%aGleY{(S0WPr2zYC6BCPFNJ9!e)LujIwn*aMq0jB=GAI
zdTOpLwB;WMKl3{<sx=J{87?DJ4FXW7TAdvpuwHbFp9>GnS_S)*RzY2qGwZF>Bt1X%
z;q~gvFjr#|G34KOt1mN|aPWjcOy6OJloid*lcc-0o`Qo4?I=@!RB#QP$XcScasRaE
zv{TxYRd}4^Gyy%Z=$}l+{24^dd#(#7U496vlTXtkUunV7{|_i$e?kN1*^=&bb#nTL
zB)fSvgBq-f!SaSOoHXZ-(C#u6Zu?cyGcE`Cp2wU0^L@x2mSa$&)`8B?FN5~nL-<gd
zVed;tl%Ho$w9d#t{M}sfr}VLGT<u!0o7_y-`#6yDv0uP|XAw@zxUs%}^NGd8PIMbQ
zhSbbxuPUR}SX27}*0kQhEpZm)yxavSgq_qztyr+Tt50SQmWJL%VYKgKym0@o7Rl1L
zrxgQ+5ruLp+#emt`aMoj*|{1->P4=w{?}T_x$Y+fewYYz^ZRkcth3mbu3Z&3b}%mV
zeos%z>yZu5vQch?B5ParS8TOfm8@T3h<EytcJFEyW&f5Yhe`&Kf;oNg8huDwcqulV
z*JBCeWQq2kF{EciJEW;pVE?8HT>3B)YFU|(GCCW+4RXP6SHkIvBQoU8n-i$1QjX?Q
z+n{2mB5Aqy8T*pzX#Arl*CO)<kXJiJe5MxDFIfqATzehyc|A#d`0_JwQwk98`R9)A
zd9i|9+z94#<e?C+t%6~NvBK9c+U)JLSE6^-WpKXNiW&v%g}`}}$^6Cqd}k`l{58_4
zeQg&?zSsdST8>0(nhIHRWC*<oQeZc;2aDvmV`#)e?m!qu&K)fmnosPeWvg9ThKTQe
z(-y%QS!EVBBasxpiNPI`aYQv|A&KhIBvsx5@mXez-RGW(-^ppP-*+BD;Cg-*J()uk
z^ul2+p8-of>;VJ0acKAWCpFmAfl=G{RGsTL!9{c4!q6@DMEF~P#bGfJT<pYd559_D
zC92TihAG+pU^NU>Zh$Y#eVOTOc~&Y>Ag&ufgxpbf!Nvi<vHiX;GxWU*5>ix{;`JS)
zvifPe%vO;7oCIc1%vpo`X%boROkjNWb4C6o{!KiN*S=0*Y~KpvW*3StRd>UR#GyFX
zcMz$Y9)^YrXEBA(po;#B5+Y=?$e*8=aGdUL^10QR+3H!~ygm!|c<dndYGp2|j^&@z
z%`z-6V-MMK`jsGc%86cm<psI%xwygP4Zdyr5C5j;;FeDZQJQyd#*cm`7zgKxL#8jn
zsr>An<>bL`A9p7WYul;o`!*<7v=N?4-o!sz&Y0NGGvrgIkiXCNk|^CBB!k}_oa(2M
zb()&&vB3>Ex7dmeRFHyDzXL?|Ocd-ry9;l+OR{0f|L3UJ;GIiNsC+ybEB#Wb)3;}s
z8u$t=KGfis>4j*vcL@zAVo1s0{vaJij4vfvve%qRTq!||%O|LchcmcH_Tr|U0!y)r
zB;EV##RF5*V6o(EjMdj5Q#JM3u%Dhh>wZblO66>^1QBSyb7lK7M-jc+yY!M<I{x=a
zk3=0#1!KRV<YUS+Odd3xZ89E*4$H={#qXbExn2&5_5Fwsa^9g#QV8Uo@f6&2p1Te`
zRRWiGC7|u>r;xJRnJLJYpxfEiRBlQJDtlc+n{lr&T;s7I45_7$ZjK}}FTGgHpc>JM
z$iralvzYfmIh(R89bgJ)fpI6KL5d4DOnyj*T`0!6v`UnD*p%n&xSL$Uf+gwQhx6Nv
z&~W%EbbonEC<{;|Up)?x;T~5Y+<67|@%_{MC5kBZJ%yB;BHBGWLG?V$(M@Im`8Mbm
zon|7!17Fn0z(2OIg?D6!%`4`<;Xbs=5#tTBdVDZ10;GmIfZgjvm>0Q*6mSMa@i;xE
zUZYKxS;fPlPF=!|?4pa)=OTOj4V>OBCv|PyrR%K1MDzP_a=#j}T$_XYoB6%#&rFmW
zc!KX(7nAO@mO{%tBi5`J08u_=RZU-H*{hFIY`XL?rlnX5#h)xWZ}B4DQ(Fm<+{aM8
zu@mX;98B2q6f-Rq$)w$!@2;88=f76$ee^T#ote+RBulXgkG{aV`|mhI!k_HASOOvA
zGRZ}+GSppaNE&5kfmdV`Y+V-$r4s4bvUdfghhD*y#(S{otr~NgwjRx!3TaNnWnAsn
z31;sFvT$%aBpka=6J6Bc&DDjxzcGUvyojgtObPUc>#?ldouce(J~*)ZJjh3N(2{lh
zNwfvQr|v#jzp`bq+^5oM&;$nB?Ly4INvtt)G0}VoBxOrG#(VUNBtlPIjJdD~BtkY-
z#ZL8ruGkae7I|qFUSo<9(~hGF?>0D(X~Ks#k=VD-40JjV!u@$xEVief#@8kY9^ngN
z&6iBPGZ)Z1{VUG3P$vl|`Y?Kd8L@Ug1dio5;6|nv`6q9K+YfEVEBqX)_q!Kg{5Kso
zHxI^Be>!p1A~W1rWsPSp<kJTYzc9jL1lxO%=VtyY6Wu5AIP-cLHjU!FkDz^c+{u>3
zj~~H`JRQ(gvw-~id75-+$bj<JN#x4~2|85Li*;Xp3vNs2&=mJS+`pHLyGuutcVpvV
zw%$ujxzr{|41Zh|#?O_%gKmQMpM_vE<15BKa=`q%bI@HCY0wTnH&-2qJ<;~;`TN<h
zX;3AWN(P7}cDuV;crK+1Wz$G>XgGww{)uN5xi_yb6}(>U!LQ9aWYLY^l%9$O&$C^i
zIVl2m{0l{sGxhlR#}c}p7d0%zIpV<eT6AXV9MY+_8hb;n!P&JFRpTZDhxnnyr04MM
zh6ek__g%4)&ZJ^xE$Y5^W-&ii*oM*P(L*IbNZj~G6gnvjL0E`}|0LM)sk>3`PBpqX
zjbif-A48j>34E`)TC}-&3+Eh#i2JYbuH5}k7@EEhb9F3H%2|UMh7KbZJhM`p=!{yD
zJMrygWxC{(9|?Feh)6#+V3UrEaoOafRLS=ewwafsL-<g<`(zF_<a`ru!X~m|V*-ZH
zQlrlCmmnr=Egc;pN%D>lppAukC_Z%@JLJ>w#D+r9KWW2^=}f_OBcG*R_9y-KRM^xu
zBbImO2c~pu^4y>iiR<pbh85j3WY`4qQC5{ab+BaWho-Vu)$6d-bOD)p<t(+o;0Pyu
zm!j!v-cLNT8{5t&ii16L$Yna4xK{g;PrkR{VoDs?CftXJ@w|&aG7!yw3?Wrk71-mT
z!>WypSU{QshCb%%&l3|^L?`#sjJAUl%1U@PnKO1HdxUR)BQSAK0(rGbi?fn~(KJk+
z?RPB0$V<^Isi_>c7%GvWF6VH?dPx>v%DMH|pWwEBNup*@42cH~(ND?`V{OdPX7U0o
z$qi(Q$Hri3^(CwuI28sfnBeTgYhmo5F4PU}7Ys_D;J*n=&~nyj_Fh7hb#Fa@3+X2a
zi+4urN3PVis~OKHeMjo_4;wT(sB;(3L{xf!iIW3)kn;`GTMW?R5Fv3%8Z2<M2iqK7
zj-A%2oSQ`9)Zg!P@|sAT6YoN{8tH+`mof-+tH-{=Z&;XFLF`7f<IG<lamD{X_k7-`
z`)nx*O4^9CTJ+e7Z6ENFpBW2t&l6(jXb}6IJJ7xK9PNwhrV+FdXR1cg8FtR>t@j$z
zGIj`Y)G=jJgZuG#x*6~F2SdF56g;%p1v4Ah@_nBX*|kNFyvuGE)ts{8E@dO8ww~aj
z^#gF>wUhYHawz_rI1}P>D(UwLG343um+(VUjpb|U(=7SH*!#x;56T_J@brg}{6v~H
z+PX2@oAI!q?g`yH!;)-WCPf@PYv|v=LY(VfAY8lT$T|}L(BY$ggA`{u{g^4q_OH7I
zcN+Uddat=7Q3=F5rzF7aG0(7dn*me1%za~5cERiAJmTJ_gch-GG|yTS4h<baBGe*;
zyZ-8I)9p-Bx@SFB)=1F`V+)pZwM5ic)`og^!`Op@^TJgxRrd6GJ+ARKWaZ)SFuA~%
zf0puH4A1SsZf#B(Tf_Zx`BF6PbPRRdYL82$13=>9s7uQ}`k;@-LP1VR3`t{+n45Ad
z{7|gHXzj<a!09qt4PJ;*hmBaxUliw@<NSpY=A`yQH9qP44dYiRvL|=$;&b;6qVo?2
zvhW^T=CI=cX*CHUK@%(B`MCmgQhW+#TRwvI$8ureMLm4ceFkUCJ`xOMi?Ppgk7&!5
z4y4m2quW$Xx~y{&8P0dA_JihP^WkNzeAHb$H?SY_^nB_0(9Q5!wiE*;(?lWXl7)?&
zMNzY)!?o#Ff>4v}BV0DQgS{SxjPN^L=c`*N;btQ)`SDfwl>Aq4*D{5%`mtpB-x2J!
zaVFS1ib%od7pO4l4?4V%BbVz(u<P2@+(j`7<58Y08@QiyL=uSUl?n*ls3Df!KZZr@
zjYFNze>nT)A&_)?jH>!$NhiM#yz^Ut53)4q`_3ishm0XAU5_wS=Df&1NC$#Oa5t2X
zG1+p-0vZ3^p8tG>Gq!~h^OYB&yyzrs5Bws`6l=54=ejKXsF^TkeG*R1Sk8HzM(mth
zJ0^7xV1?y~tK<my+%19aiV`einiShULyL`Edj#hUHexa(y@<NL8fy$uVclYRvK%eZ
z-|H&IhfE{~bF^4_^c<4FXVDXn>5?0Krd#@1ORQ9>%<S$tk%|7JXrf&ob(>a3z4ptK
zx>-$BKGFx1rzOJCKZBWqrzg{UsE<#^JjK?mpMv(UMqG1yA6gwegzSY5Ro@aumaB%q
zteUk1z68>nr?bd+*%`!Crj>5wyt3m)_0(!;2d>o45-v}?56$!zU6wZ+%O8lbt;?I}
z3p`)FZ#Xe+sR7sPa?G%!5Nyx9gSxT1(eAdUIP3FMIBs$uF9mksn;8pn#R*SlJodV%
zw?k8;=yC*_vt*g^yIFXrz=D+z`a*5ysWac_>BJ)X8lCN(3{QMStmwE3q+D1}KJIlS
zCG?RX)Ak!{C)`E1wR^F419ukJ8?rL(dxHM}DK@pB8cX8*u=}$u$Q>VrUFQx!!}CDa
zbH|Ul#r#D3ox#j++zg^}^d;(#t%h$pPGD{IK{$Lw3p`FeM7z1lcr9ii>$*HvP#dqr
znr_U2yxKv`Enb=2`>8;qccx=ePcUKoq=Zq|HCgMgWkTD&$?T0sAp{pi3HrM?5p$l$
z{MQzT(Fb+l#f@CN(*XGS^#p8sl_2z=?S+P-d?CihihUnn2;u$ZLgm`ig7k$3KIgYX
zh2;a;6P^?I*Pcuoc4d+mFq>_=un#ZqlVz(vRZ$uhMf}X<iS~Ch*4zFcD*2AUQ{fWC
z?_(2Ywv5K_mNP+QTLJeb%af4j>O}8|HO;#x7SdD;VKQg8?ACdP*3-6NvDqP_<LJtA
z%MBU+UWbQ&eirh_AB1mV7qPH0fUus?#Ae=ph!6gPHSSBqN}VUc{zi_FZDoM}zSrZl
zhawg-XFpgq`Qd4&A5ie?xM)wT2iq^h?*{!3G0J%`mWFI1LH5o>-YbpgQaz!5^+Qa!
z^A&sRHsVjoY4~o9Ix8PKo3+LLa~0}pAWK742pO^mzirGUZEv`nUDk+62Be}|;cnQn
zAOpG=+6&T0O&G5W3W1ZesQ2_d^dH+sABl}IyyLgX;8h)zTs|c<zFbC1<=VxP^9yKE
zlqK<PFB1Bfoe*ZESE7`9Jw5y19Fp&sik&fC(5*g7*cnxTYH4ohJS$u94_e3b1pCOL
z`Imrh$%FkZjUa7(0ZW}0lb#3lDAe;FagZhR9cscV&JJLz50+t5v?;r-(2M4`bl9P7
zDnxs$9y8qi4$SFpYQCldFDWDnMmN2gqK6-M6&)dO<+Ygjg%{e5I!Dw$sX?(q2)32z
zh;L4RfPurRh{w}~r2B0KXCoFv@o`VAiv5V+jz-eboQt?5Q32hZb&0RBIf?Ad2GxTS
zP*T4i6OAXa@7uU*G20crXYVAH(_Fx6U<)pOREP_P>yt}s-h<7(5t!iJ51GxkaeH_f
zex8>N1?_*SelWr%jZ3s}YXVLUISLou4v|pjQ^MQ$eeiaRG4bfv!O?%$li;{ESUK06
z>Ce$c3wg2d?d3qWcE%Z;%Gt7e?D#pXc`H7@{R1VARzSy?UufZ3L?%A3MRn!hqL7$)
zbjjpdys9%I;~8RT-s=tng74suSK6p@RDtEMbAyNp@=P+*1GlAqL!Fw@*mia(o942G
z+_Wl0^It9aJ}ZT~Nkl;~XY7@nR}vQf3xPXN1wrFF_g*Gcfq3$16iG>f!plkIz>x<)
zr$ymADLzw_ab!*J{z71)GD-74k24F8V9;AJEKcQ&iAmaULaqf}J^6QBp%PvFl$cXQ
z9tK$^;EGk2Yz4V4L<Q}x(pG85M$J5Qo9#^cG&$F-e>msM_F_@&Lvh=SoucAR%_0^$
zn5-CdOK7cc7aFE3u$rPhSad^+xa}wx_CHuf4w|XJ>}P+#$=D5p%Jf-mYbxYCcmf5+
ze`&<@Y2@X%8*n~7fxEXCU}1$kxN%2p>^W&7p*B?<dTbt_4ctS=nVC?fv6Y<2JCAAS
z8BJ){U}{l|p{9fHYx8Q+@VN{-Ga{I*=&k^B6?^7(Xrm}3-~((rX2}*-c2Mt|vB>rj
zmT{JIWk#1#|Fb^mwe%IH1QGJNq!+4Z7t=oJ=X@t}7MD4Rg;76T*pBbZiJ99Kj0-M<
z9CbC0YptT4M>&sPUWv3-RifDiJF=(XF)luH3*LBt#T)w<;Sp43`>!k0(B;`;AI@cJ
z$yP+-HiJwZeH@)^c9SJnq}Z-Kybs?0R~#1+MQhz&;itN1!jWG;5$EQieQG}Xym~L1
z-`Ifr7mQ_D-tlC~(`>XDA+VZ^Xx^8$7g~L6SeF=BQRhsQu$FZF+EFiPTM1-ouqif8
zxi7RIt3mezgPGykZKT!HPyD|156>eGMz7CzVR>-`r2BAR#)gk{T4MqZonDXU$F9YO
z|9i$xtFdb;@fa=&hQ)T0xP8%m&@xtF5$=Y}H-8T*J33-ugB6K$sX}M3HqHq*1#|b+
z?D)uY^i=Rg^wnKLdNrDHg6V1AJITYnPr~rS$6>5#uLLVR^BD>`H^Ii}I)3*C<{kVC
zK1Ylv+g%%|^Nkwue>n<d#oLLH{K$r0)+`}G;|8->>3$kGLk=1hMv;c@^Y~=LZjyY|
z6^B)~(#jdLgoxjBQE~fQyeFy<e&q4Iu8S*s*fJ7zYDcm?-@VZLP!_swn1KrQA23d&
z#HN_L;p&D3#HnN~+2kO{bPV!nuEj_avD%dNb9UM!!#$kIE=kpfe+TR42%7LJ3H>Vm
z;7a>!;S@a!d-C|ZzUB>teoUZecXkTeZc^m0AMaQA^z+=&B+^9IkeY8(Svc=$BwHDh
zfeGhfsKQc`d_kKZK-Gxrn=!2Gg)>A}T*bjB*YRAV7uei%#_0i)SegG$7>`5Q)d{0m
z?ZgI59cRZ3lf2li(o!U?7Njvan0(Pz!o_6+nX7L$I%k(+p>@BYw6GS}r|GlB*Y44!
zHTC#B!xKEOSg^tZBbE}Y%!)oN5-j)HvgOJ@gacVCNI(+rD7B8m(08$Dvo#-HZ5Y6N
zkq2GlVgoSSFNQQf8o=Z#W(kp|63i}Cnd!A$z>YD|0#$vDm1K?3@S_D24w$g5g@kBE
zzs5JIvA9f4iv8G`LgrO(<euZ5!tDK<F+%bf)@(3kCw{)A6RTZV^95_>)T6{^DJA3U
zWr<i;cngwucG7R|X*k!%89xpk%RVj7$Fk}FLBilKXq;_G!(~#$KHZAsXk#UWORB??
zrqTHNumE#ar9rbRf!JLdDV}ZXOE9Ne=w5Szw6rOR7V+m^MraE*hN+R<ah|Mit~=bA
z@D2Qfq?peV8RC6;DKxHK2T{`R@uBArY^b7wqi7fzHeHtuQMpFP1j>;<`{7tuGmxy|
z?!pK8bA|Vr`b>TQ1d{rh@h=@`i}}tbmsvgrI>`|Wha3!1s2ALxj$%FuLs^v4Td~*h
z`JA1;8e%@y!MkmvFoKQ6zMwW*>^hyaJqfSMklqbdlBL4)YCqE9S_Y4v9mW14U0iYX
z5%qEVPJPxNpb{CCm4ihWxtDktjEsMdL4T9+%`6*Y{;U_JCV$5)%UiJSx;jfTmS(wc
zM!=j`66A7(Idi`UP%!HuMCINRy?y)?6?u<J!uhs~*#Zmp#>xk;5o4BMdlwTYJr=st
zCjq#-61Aq=xFYkP_<p3DaDBWYjmZ8+Pq;*p;?<H&yT=sG6GjkvEDOG#bzo}!vtVZK
zIP&gA8orL%P2=z8(OLK8*ejmbb{pz1&X?bhijh&IT<I`~iZ@_MQl~iaRu~R$9>|P^
zU$}&KByOe}lD}1!%;E4TGA3+3?tY=nE-Y4t5Xm(}>OeAH*{((E)f4H_vIDRuD+@|r
z4`+i?FVULFk*>4f`jMM%X@Z56H0xU$j|<YL!u#P?IOSs-M7;ijpCn%4@i_}n@q-rA
zng13R_rAw7|6b8!$DRvC3%DCAa3L8qG7U78`TX2(0y{{w$c3q9kbKG&gT}8QBYQ4^
zUi3JqzquK{>l0Eo^Der2E3$i=J_??xN~GgvA>2PJ!)g-8vGE4G1YNiP^WnchgW@Ds
z!5m?GYYnK+GZsxsbb&xU|H>}iQ*_M3^BDI0w&>59p)6AT0hG24Vh;cBa;SQUD({|0
zhbv9+qq++*+Z614%lZ9FihU9truv4NIB&~wd{vu8!>=2|my^M`0CwZNtJ2K(yfw_}
z+Kb5=8sz;pDVD@{l6B8-2+`4QEOnM8@d}>_!*dUU>z*l)bnG_0RbvEKwq>BipgORs
zk3f8#M{cgZBGkH1WvxYOL@zrP^Ipyuo@8o4;~QWtpLzuLKZ?#g8q2Q>!^oVOBxEQl
zl29S<vrig-k~C?OBu!FDk|ar#gd`-HLXsv)CA?>!s7Q)bny9E~kV>VJsPBA#TdT$D
zdEe)pz3=<F{@RnkCHky5GXu((O=B6S#P~QnADdru->K;;s^sp(B3|?FL?vl9t;-B6
zG%kzn4J$;q;w};w+nr#aVIU~APQ{~hQ^4Z-ZDGX$YfS7vO~SP^u)<f8-OKxmqvzLQ
z+CNWnbJ`)%xf&%%`_zaxcFZA(1M@Mdp$ub8c9N)*1ED?NfK9*Gg08Dd;8deA>*%=;
zW&=im-0!34w|k$^U0e)5tMmoOe6{jEg?Ox6;6&Q~oP-*C?$}f9fF8}sOrm|KFc%LB
zCG(`2v-D{8qf21bj`~b7Qowl0^DwOBA&xLtVYW#<m}h^SPPFF^h)t4Y#EuA}FKWdE
znkBZEQ4?bN^@&L)&q}s!f^n~th}EWg@z#n8dhWX!&Hk51VaR?#BKp8pMXw*|;2DZq
z536ZYh6hS7P-TIc{I1B^Ht&W_#^vWz>F(&k#2_^QUyYSyQ*N&!-RqpOVi@9sK@-S>
zZ>pp)<&$6|Re+j$*QxF6ADFe~Gk8vzMXX%@;k3bR)M`R6opEh4o3&~+D^Q$;TcWG^
z%-sZ6mwlw!xjKwB@O|-Db-FS16AT$Xm7FhK1Xds8A*|a6lwP?rmqAYKOqMlUVJ86{
zN2SrL)=}hOqafrpFUP)=ENm?|V1+S~I6KCXNxQU*wC$ux)$btmyZb<Jzc`gijd%(<
z4-Ltrty9S6XZy(HkPwvD+9b5Ta-^%CM+m<9$)x<!ApHDU8U8-BK+BuMm~f$ipRs3?
zw-;v<C+|PBVE+lR<FLoL>OVQ+f9nxW*3lx*b?h);_B8aryp38Nlfygu0y})2-|@t%
ztVFLDr&!6dra=w(F!MCbjj2ShtVNJ<&4R~K)!^=D?!)GL@zR^>q{eJDs~+`A=#3i;
z+0#{p6j6^5vC{#^xjB)5!Fx!;q0Ll3Yydm6=r_(Vw<FSJpRlLPk$6jfz@92Ax^-*;
z-)%ghm6@F9QdEgy>w`$n<Mkx@>1q1D^bp!VO@Xy<ioo%xA<Y@G5w8xHBhU55(bk|N
zbpOZ)=o0xJcR!p)#_LwXW9JxLSe-|8*FS=<x-*Eda0FA6{03f!!$jYLD(JSED~O*>
z9KQ=4#NC;bIoB}~Tpio!D2WTWFX<c}=g`fGH*@fEXB}O>dpES)I)!U*wnC)Jd${u7
zU3$(XLR`z=wX+V6W=Rig(Kc`y>3pq-)obD?`aQ-3aXO!&ZAOU+3&k}_fiR$^7G0#<
z1g|nh@sP_t)LqVxRYpDo*_S&&+r^Rj6vyM)Pd;e5aVe4Ad<dkUXoGc83$@hoz~In)
zIO@EC*jA}Aot!5ke?MKeL3ax}=IWN8zT*qaVvUIGG((m$zzLfgoypBWU1sTUk4^|P
zV)WoH46_YD>slrBN`5CaD>QJYw;k)+5`f92wz$)B0;w;~qB(gdpu;<c{C#T3O4MCx
z=N~8N;yd!KEjRGn0#|6+-$|{cWl;IuQ0Caj8Df5Fr2VEYVbya$kLt5t!wAgzR*UCC
zkD=3zG^#RKjk*1si07#^Icax?X9SPHN13ItV^9oq^8DAyG9&Vj&&GxZ%P{E=N6-ms
zXnbTiw#IQj`-yanH;jPK^fFaBugJD#4#Tkdf1!S#9c)N+z);7H^uz^OmTGc`>e!72
z&puaHc=bO2o?GOQB3%j{%N&SXu?w8xzvJukxq>MkAvaukx3qUIS#;Q!#L>g(Jb-&l
zW>kribmNJmLqDG{&th|IcEYQIN!WUNE6tc?g9(cznN`1!XsT%}8M8r^_!OCtJUhPc
zf8T<yYy5E9?Y9sTnE=@{)yjK)7YXl=>yi@vG@RG+5-o~Ka96h+`Pp&=lEaJX@z(pO
z==TOw4l}H;&lg^{)lkhZquD9N3OJP2fVJr>A>V5q^M0evED8<5A?p^V{#=hGYbUX$
zmJyg$$5|{_WSHaW1i`7&0D|gn;QO=}C^522B&6?!C7yv09MCS5c1n?CQiWHA<2d2(
zCJb?SOMCu>gYEZX2x$rd$7z-%`-?Bm`~D2$R!Fk^>q7<8mIY9ua~GeEIZL1W*pc@M
zHr!Ft4D(KnAc~S_;ewYbv$tyyxo(Og&DN7}l=d#({i=XA_d2Yo`i)+5W>XdZ?4Hoh
z<m|1tSUIE@mQI>MEUwHTr4vk;Vp#x@m@p9QCCf3M{iH&nD^7OKA>sR<p;NK0FnP*X
z)Vag|&F&<k@T?g9j&(qr|1@Dv;}`n(&n2OxzZ(726!EP^8E)SD15%F3GylJfiRgSR
zc3OthUxzk<d$tl&2)hBsNzzQZ`v&g+HHlrdyNvtIO_-jx9)atpK<#9d$hC4lVT*>5
z#qPU7W!(+f(Y*)nH8<gN<s94`S^~<cy6p6koA_~cGe2wpgEEg<<jARo<VKY-X~-Xg
ztB;Onf5*JW#go2b{jd?t;AJ$V^344-*&IyzLE)6#7@{|YyCr9z6~=cJ!gkksylmGg
zx&cb?)I*k?`k0P7Di?`==Ol8;nD?8<<_IlA)tSldPp~R~AHIq#0?$xmqNOQCHb-*D
z<lFBUasC1Xt-Fa+vN~vSDTUAri6|c)jmZYd^y~b^SWqfJ%F%HsctzsGP1m`b`UK`$
zw&D~gJ}a9iu&BRrr1?V%KDf0E-+xB%8J#H33#>$?fDBYI_zGWr+c5KC8CFQVLJ4Q5
z(ln(+Ja9uFkLfQKVp^UF8Ee)OKSd=rj7`QTOLw5Reja(#+XODBm00BSHmK(8_=Y)l
zpmUSEZGL{FL&sc&!z+*A;Wj>}`4~i6vd2N9VidI2#fckF>_n~eoX0ov2tGK|39omh
zfsy+l=H@hjJs!7<tX=m8{9kQjrEPH}FXAFPk9~_f)wP6`@xw^Tto@uXY%Q=^#kgOl
zorXp9(ce!eAYGR!%s6urJGdjZ>F@_M*0>0<BV$mUs|Ce7-h#f#B9ZSpic9WV;+5DM
zy0(2J$Tf$9bZ<7u<fp-pJ4T?dIG+7XJcPOYS@p5&g4iilh@MNB<eS;}Fy<8T`YFSD
zs~d6Cndd@R_CPi*zaC%TH-$Oc8qBM%QE2><fIoNI3!8WkKsEh6c3P*1#$5PCoh3^|
z3CYEH+AtSJ><`6_3PZ6$x=pn5Zzx&+uN37{XS2CwZP>s21gMUg3GUC7nU`9oXkSnQ
z)^<^NU#m*$gFV?zySKQ8XXv)vVL06E8>%<ikc{u+$n(XmwC0}-8zEnYaosf-tqm;a
z&2muuz`LE}z0l9fmbLx=Ubj<|{JPQ(m(-p^s>f)Qcvl1)@9m@iU_JpgDRTJAaQ0O8
z8_GMJqNd%(DAH<%n;RzJ=ck0+xPA=Z)(v0<Z#2pBpUv1cXBOPKJs5wLzCs5zZ&2E;
zEv%n8kvw{7L5_@9Vdr9BfYFJCV$Dn?mZJBVdMPf36y8gC9j{M_mKf*cXyINXQ<Q(X
ziS)l*hPxMbz||Zj=HKYbqGYb%qvQ_!dqop;4YK%sG6q(>oQ+F^ZMiF+`$Amf$!<$y
zIQ&taI1JRrh$CTmcX>HygpVVxc_Wy!`9<u0(TE#oFJp^RFXJWeuT=X+G3-<90Jlx*
z;Quy{_FP>69c}}0YzNO#^St2Yp?Mgfq)Q6Y<ye{AIp~!?B@|}0gL(T3T(bE)Jq{H@
z+}TSg@glQ)m}fRg^z0P&=9xh7t{D1X_EBs}trnCEZVEL=dcbv+GAX$+7z34(xId)_
zmG{aMwS%#uU-G(S=Tz=}{LqK_x4%-m*|)*r_*fPme-z*A9fxm)A5itcbYeQho=oxg
zU<Cz(*)Q8>@r`>AP&FkB&iP2OwS|VPc(OJdHA+I{m_Hn4w#B1qv=3X{(t`0LocKN>
zSRA$gCst&C6}k0#VFlQt#*z_CyG0xR^yk2vT|kbuZ6@+PT^O!2h4%3-so$YWj8+rT
z>)i91(Kdj@eNiJ1I*#D{jme;_qRCp_WSHE5sh}0~O9<SVE0mOQN79L0azV|A)UP@M
zmpz(<!Aj+zbNM^iUEPB*7z9m|j)T|nU2u8TXb5_A7M+%?LgfZ2w9@PrB+~AJc32-S
zTsVe3TXh=u&N?AHDESSqULS+S0q?M8<36654MLZfJB0Ba_K^P3mb*M2pn}#yzN=GV
z_T}#EPiGSOxqSj#6_SM0##I7o)o1oi%Sn>O3KVHtknZ)%*@28ejCd%Eahg1>_tlX7
z7xf0!1%Va!mO+@k6pQLlhrU5|`2LYP)^v7axal&C%y<Q*i^su^yPCv&kSwTqCX2&I
zAI9Uey3x_YiTNLKWG5_c@v{k^YrBtPMh?m%ui7`9f2T~4KYR1nb5C^gZ~8Q09QS+P
zf`t$5S#I1unEhck7D=1a`B%?knw2YgXlRIs7I21If-80tZ!+veJ<K^K7Ch74VPBy(
z`R{u)UY;0*ug!Lnn&Aw~U*E?m#ZzHegbdT!vJ1W1hk&4Qm5v&BfP^iRgV<1gY?~AW
z-TTzwpLsq=$0-Yva@uUSngo*`MlthW5$=1ePX6g1r^eD1Xv8_yYi5q-Of^ktzAb{K
zcTZuNn;Qu{z6pJGG+EpDNBsZzC%t;?Fb)%^;HJVSaHw<z90}K$D6P8?`%3wpB-MZf
zPF{txdmd56ef{+0u|e!W!~^&_{}i0xD~pq^eW5<Yl*!F8B1$T{m_2AFae4GhoWFNI
zdn6phn|4KD+@#D7YiY9$2i(zdr>Dc$nHsF5H65ZnOfd7(Ge~kuN3Vz|QH511J&|P1
z<YNwSuF!6Wt&{bcRc|L9FLY3*;{h>k!`X|e8&SGViuKI64<k+_W0Q*nsW@8iP?UQ@
zP>5KHhl*E{KAs!cKGTU*M>Jwq)MSvps7y4b$-%NEY1})TK|aPuz{@RKtVr&eP(61b
zlYc?Q8~-hUR%aiFF8NAado~N7K8h7Kth@j{>+gt84ZDHozUq<N-=YX68Ig}GRIz{l
zNaj7?lE4R5rZwOyec578Ja;?b?KB1_gD*qA_cZc(>tZs^;1C>_?!_XL1ngsLIh*mV
zNYhS=v=(LKphyLFfalOOZHlR%pe@Xjc3=?+=IrK~RP?t|V%vS>$We0zq7=1*rk*ty
zWcl5hMqYrL*u%8v`)SnIx1qW&uP|=5A^F4EmDc>-d2pOH8}+J4Xx(s>eCXxe15-d-
zn{H5%AILt~Mc@e9hRZ%=!2BwEjHps)_G{wF*aQdgpOGVuJ!g*fW%6ufE6+<O?uRpx
z%_yHKA|@fh=v%xU0}{)G$@=H;Ua3IbYwXxtj~qH|+<j8W&)vT68T2Oa|2cRwdUKo&
zyPBc_eOU)^_d6d_z4#SpB<|;(%VET8<5Fts+eF7_MA1<qK5MDdWI7oQw4l9O@H!cU
zyOS%y{a3S)Gg20fYr=8M8cnvb+!$RQmte{>ML0R+C^^%aO#B=x1)C>6P|GfmjZdqo
zkCGYbENjO4fL6>9nUJS@t*HG%V<BO>xA>#Ocf6!!LN*R5z|#3=V36hsdi3r-c#q5Q
zq2(Gv{~N(OPbaWmHxdGOl?nA1_<3#JHj+1Rn$R_>5<LF+Lc^6^ShLuJU0(B#TA94W
z-NVyi-br1ytj$!|s%AnKHY|dTnPEcCWFSUMN3g+n`7`}=21=-HD;qZQ9abD2A*9NV
z6^hIB$Xrt+5yw2BgmfZ4e_(*mo6C?COeZx#yNP{(J#(A4oeCPWaPGL#LgS&^5N<w1
ze5h8A1ZTwI%))Sz_VW-93Q9pY18?&3_#vzvC(TaYH)WP;vGAZoooL-ak$99o%kD1|
z7r3~gcIZ~2@RvLJ^F{-`POd<koI3PY@5Vn%${>H_XtwEZ9;`g037@AJvd~GfbWXx6
z7=CZ6AQ7WUY}c1#=EY9*dz25gCwGzp$r-Ts!&UH#noHy?3b5O}2ft~l;Oto=u~Sl?
z`{K;WEw|hF<%b5j^w@}n2ka~FxXpuqoAgNI(_w<ys=KI>lnR%p-Vj_1A7aE9KC9t8
zr^QC6u;Plm!z@0pvCEmvoL_sQ{IvOk{IAC}ZBQ7rwrY!hNX$jW0b_8i62n<<|I$$p
z4+s}>a`E1Unb^2+1{S3$v7j~`$TjN5kdtNTWj+|ENQhX_lSIK^_guM-6Oj9yp%s_v
zh-Q+iLY2b+lDaaD4mxpL_*@YWGn~Dl-2Mo4cy?R36K@6`8=iovzC2he)!>xnudvJA
z9bTGigT&9t4voG+)b!{h#0o-08jo<+6MeGx9-o=bGr-m*A82UiVr-u8jytX9v0aBh
zg29A5Fw0%WLjTSbcGs$4*OD~R#4F$M>Ah6$nK(iIW$TmYQ;dky#Z>MiF2Xox3pQ$l
z6SgKg!xj%K7MiVv>17Hm<|#J=nI+<QT@iaQa4R@-u5;*^Fw*$49mBFYo9m(@iS;vN
z2KTIp-|%#De47S4RhmJK_8H)W4e4O(Xo{=X4x`2wC0Xz6eK=hCD7w{^)4Lx!alW!Q
z*`G3qNza)EcfVLO)gM_z{BJ7$vR;7?I|lHM)E;QttIV#gc>!<u&et(*7ipWpeUray
zQ0v((^qn=3wbe;7Vr9&fX9l3%qu*H2lM1fS>%nW|4zZo(Mq+!Ql|Jo=N3moyPMI4^
z{*+I^jn9=x$EdZ`nDb<MZw5mDKr{3i=uh`Io&fE#Oj^5mA7n@khrk~<#qRE7s72iw
zjBz<h6V7+iuE)W2rLPShp-C8iLI*{ci!tDf6YGxpi*m}<7^;<m83Wd#zHJ<ITpb2#
z3#(CTbs6q$ejvo48bCl7{>YCZ19^6E{Jl0fx6z9EC#_`<y|q}%6C?U+bOO|gB~jmL
z1wJUwrKa2)7t`H|qwjGy@sImBb#Dun#^mDb&@9eWJ1!cy;RqD7oAkmHQ&MgC6McLS
zLcy6?w0LO?6y&X@k9EU{M7M{%=XdV-njVI6o7_o`(@D~^bS`V^e*zlU=dyZrHFDAM
z0?K6F!-*qhSw!$qbduPL6Fj1c{IjX3&-cBtQ;(v2i7YA@UZh21Jy`gasUn?QQ`sw>
z<AThSbR0W9g?5gm7$4GztAbKMdu$_avp9f5#-9|5X#(sS>qfd>Sd$b#4=~V>X65<)
zSa;c;DIR@~%M(_>j7T+7yKFz{Gl~WAL3cPOeIBJe#TfLdi&|X`=PaZH!rF(=(P*+f
zvpRYWa~E#HBg?ZPZQWR&BUEB*eQaTq#9-))FvUG9hq4FZ672bw>6mU>1Vx4ugiDWF
z@Pecj`*pSvRe}#-@c(DFx+d6G;wgCcyujCMUs3tn3T(;_-v799fOJG%haZ{)K+3|2
z#q4<rZblBIdqFKkY#xK*yExKQWeR+n!QVt1EojEb^SI=b0SzCX$={2~#HAshdoHrD
zv*ssMjQTh6XHzg9{8NSb(%12Ol>v^>Q(zmf%d@$&hT_W}fAZE}6|%z|A;r>$P4V(3
z8yz+ZUCJA9)%aK_1rrk5E5nSF29x&mDyZ=ufoq=`kP0;`A>K@Z+0>=crr!_H%9b#@
zE0!eBb|lVw>ch5gK8R7%7Ghw}LcIOli2N`a!Qzq!qEv(iXFY_|w0W+iTy`k9uQG;}
z!<zAGhYrNtGKAG~rp%<Z2!H;#huJ4j2;qG*L_WWIVVK`2%<oobJlRH$jpuh8Icaux
z>R&8(C;;8YdAKFU8i(tyBHwkkqx9P)xK<~IY~3^kvgQ`zC(qlUvSI|AB0rxT34MiW
zAV)T=e-8$Ct8j+049N<92RG;3!EK+f!<~q^q+_uW`@8-H_Qd{%16BWFo2D%5>9v69
ziT2F)PZ|^}RYHJ74?oMx;;xuUtl_-^*vQnv6erG+r1R;t$-Pi?UzR!i=Ya|vCy*Ue
z20%^dSL|($BksY&p--tB435;|-OuIlRP8tz><lLPYuCY%s#KyfZ7>_Q_BMU4?M60j
zj6u~#C+33=!mcm-(J)1gBv%Gtw1X}EnR`HR`qoHS8Vn%Mqt)5@$~WkL%mX_@d3X46
zk>K2NfVF;IM?Y=fjiEEXa~=CaQj)uyB%ab?R$56&zZ>F-rDiNK&j`$H0$})pW3Xn>
zVOr*;z=m(Efx#XR=pCC}*ck1C*Y6Zjp;be${gHw0H>*XqxnginT@14g58<vSvh4Jj
zX+&6*Og#J;6gGdQ?^7nw=Q6)B=%GGwQQFJ1Iva68D(}}6sgkPiwe<Dv7_y+%iiAo^
zP!pqe=(4sY_M?@URn8D*RKxi&-U~3_&y4NQn8EC?%d!tOIXH3kQ&1Rt5o+oqX_LAJ
z++FZPC=6?+L;PnjTe)}CMtp%*a>icr+;oy_G@IVNcpTXzEwVAvkBuL90B(5fr&$-4
z68G0Z@LIWvz8+Lhrc6;L<DZ?Uc|LnUd5t9oeH+SVh4~UG-68ny?F;N%y$@E|9j2}Y
zqgeQ^{pfC=iJzC{;;1pVQ6xPHrmJ2+jo1cMyBv#u1Ua0gWlH4!TCqn#b6G`Ky2BGc
zRTge;M6$~_u`x~DO~Sb|-6sv{@*rzW2&h7(d+$)=Qw4Sgsk7fNWLU>kNhbMX8W~Zo
zj6>!<!-*e;GV8i)sGW2iGFDuqWaueeXgZs1o#Te4@2Y55^D8m!eutx<M39oNKjDVM
z2GAM13366$A{z}y38CBU*kpBOY`6IWH#M#Csf`QqI<*rl_PjyUX}&mZ^FH)+tHi3?
z+RSrQ7j$mqxto{!F}HXK>AXG*eFv_kzCUlk&x7X7!C?z(C=cW_fG6Nu_=G-S5ojLH
zGh_#gsIu{IFkdzf!yUJvM<92D<bOh!vqOpVR%0gd`X*{^Nr&{;w(R=1b<p3U&Gxei
zkTf9?ial3y$L|>8PhH4Bc{T30T~ElmPdNFq0fs0HVJ}bcyxU2{DGMhvX`3Td@%9-g
z87j~2ZzQZ?ca1nqXAe?&TVx~)yY{@pxsQKhu>Un!`}8%KyisS>$*u72#c1}vT@F^C
zx{c?2Qn|OrRrtNP9!_Wd63+csgQc9!dh4b;v5S!*yLC9<WZ?xc-yzSOO!~m>N-TO^
z(1nxNM<bCA<mU@9-jFxOgbS|Z(#{bqL^PJEwPax0nqpiqmb<WxrjqNmOGxn6aG~T^
z04w-6o?YMe9EGKKg<69pWXo_Rw!Um0cKUo0T(ws*bXtiA7VZ&Z6BE%x`@7J7BphVd
zO(L$p*W=ASFYs@97i5GDWZ4RjMA_NgWBzd+COqNp@hWGOh?`V;>$4=&`TSFS+2t1A
z9dw8$EQ~^b<<(Sv4WM$S92*&_!8-Vi#VIUOJY&>J3>*3y6H*4DkF^iPefmxhG|7^g
z_fCqH#;+iWs=1gqWv;NM@fz<w`;mdG*OI%T#%Q_uBWh0`$12B-Vr3h7CWqhIm(-?;
zzLySQ*`@c;(RVI<%MON>%d${o-ET-QpU$=$UxBv2m+;uAMdZMT8!(#hr?d5s2)D!^
zVA<3G<ibbpms)Jh<o3UT{r^oP$vZSy^?~2GR9At#IJpTFYum9juooVGqxkFheR?eU
z37;(w10~NCOgJe)1B34gX7O{_xB|`(QI|oZm!9GzMV{&Wvr`DXR0n%?H<5zFeZosm
z10)t!v}VOS(btl5IC9!vJU8MjNxZTHN30D3GoEK2uc^hhjZ$O%RwrT74t3I;(|`*M
zClEaBfJOBp?0O<iXFn^#Z=oYeVz&m_@FkI!KGT75btR<8Qk{;yF3sHhta0a{g%DGu
zM*<p@pmgIy2w&nVlKAiVM6D0`!V+C2a&4>+IdbGYu5sK?)|6)Q_uPJ{*e1vJ*AFDC
zl6R4keWQr^6Ek5g&$f5Ykz`qu#*r&gT`=|1IrK`BViJW1$?5h;)V;75rX=vYR%k8y
z@IF(Y*;-sbnZdK514PQ(gzS(X!9IQ&NtUiX3kfxP%>MjK-rxFEe*M5AJYBBGM*U7E
z8oq<5?WNZsp`17I+)z7qHOP|1#c2_p)3)%+{3n#nR>eOFs>Ea+_dkC<CHy@)jHtIJ
z;OE-}y%txCl21Jl!ndYSvFR^5b`j@6lEb(}DG3eMUO?|F#h5&L0Da|e%zO?x0nDkS
z?^~8~pCFK1Z`DY_zgt4I$9FV3-iz(U$8bf{2xe=+S@Yct^Rl^%JCgV2o49lB^M0aL
z!@191H^f6P$P%y1&&t!or{fg);n3whU;NZF9i-ps!aAR7=>IPW>(4#Gvle;q)KeNP
z152p)qe|5Ju?yV7Wl+mBPS|^^6AgS#a2oe9h359)pK)z4f7>wjK1_}TyvW3LGb15B
z&6sIDlpuSy%w+aBg-G%ayN8nmG02)hKHp2nis;4sXXa$+_hrOVQUR0%6P}&(Wx18B
zaiR~fu=WyAEV_(4f5xFst|3Z1KS-S9J=xzp&TUZGg@Jm6p<WBEE%=9x0cSuX=LzkK
zN}|bpKH0i@7#Wv;8{c~FAr_a7sMek?`aWHfxE|dDqq3LK=&DWRPEsZOT`LI(*AHh|
zt44ym`$qvA&*B$mz%z2G*hY>CbN^MtDR&LfjXNpit|&pNg?&(Pm9xqW0S=Bi3#i7q
zhZ{86g`^>D@K9BJJWhjnw0?vYrK&vhtxEE{5d7K-$-4dVSn75PrJmW7O8={%ZIVTX
zPT`EpQCF$Tq=BSJeg-uX6o}K33=AJg#16O3nRgS<*Usd*jv-$_O@1kAn;C-Hx>yW8
zctPYdMhArj^;mi%8!j4d#bD3zWD`n}IDYpI{nr7}4|8aOfg*dqS&p@{0~r2pwRqIr
z3FNt29<lnq0}uVQWMS{**-8CCmWJ7=W&1|((jP%)?G|I;<%7biGZw5NS%obc^A&Uy
ztwdkyl?3kzzu{`^WTN%0ncf|(Kq9M3A?SuCE7$!|{<G(dkRgl((@9g=<qgTG^idj(
z&qR@xk8fd5nU9d7{)2NuABun4Z3P3<Jn`Ere{fOGRXV#*g+ztLg06%;oE$NgmKDT;
z+Nud4<)1;W#a=_lg9?IHVi+l$sKx@%ufxyDCvb20b0N&^B3_e!j#G~3pl!S%$e!QB
zHO+1?I{7Vz@m-KJ9>pTd>)7XJ!fYj`5NB8i&mYfY!s`Wsw#x%}F~=LfWgmo!w_Sp6
zvMi)^gkW`?EDLe_2&W>h(^uzClEl`x;Jz~+L>B{bZu|s#UVRKZvcQL2KBYt~ADzNS
z!4WiMU^{0~#lyk`#9YB2{c{hC8i$zDd5i5}P(mf>4IPSc=Y9*O36c23;v9Ob6rzpA
zAQqRKi}P=(fy$9TI9G<dO-+x$<0Z<h^mqljta4!0){2n7&zuMgP6)x9{$gg1E$ZAN
zEW6r}`j612m^hT><*vlZa>lGyF&qChAH<gV>S*}o5_(;ZCL@2$M0rtxu=c9~vx=DK
zkXS53%!if39dBFKdtZeeQuW9E`4_q0?IOO*AH||S)r%`m+oHqmI`IbUy(Ef+;FPBG
zLh7ndXp+y*U`i)2bHQ42b-ESFema&(-}nT9<qfoA`9$n2kH+pldFY#Z2nyS>(fX7j
zNvmDTHb0YN{#pxI!7+WJJt++jE**+dcFyeIno&?#F3G+Rn8Ox#=3%$KG|_WWW^1Do
z$?Xd=5NbafhpASBy+j+><h-Y!+theIY!-BmyN1;t1h)810si}L$(mnz;ZLdMY^Bn7
zq0w$THe0+D2Axx6UHTJ=;N~Or8{ZRLN+u99yELFlQ;1$k6lRbN!Ddqg>7Kh5w=6zK
z)J|Omf9bPguWj<+RMAo{TDK2tzDyMs_U}d8AFr^Z!5N!}r$N>7YhdE@6fJQkvC7mS
zLktJ9@L@ZI#?;3c&}mKfK5<~47T4fM<wP>Aza4wcUV+AoohS__$&?*EoQD-j_WhM)
z*ET7!Q#${FTB;&@csP?h+Ov$bhOH5=`@(zgzMKmodC=jtrvZse8_Yg|4zb!?MWtOo
zLr$PJQ@yFfZrxXBcUJBu`()Nb%Z@+z*EN!}dz+yC_(1gZj3J?iE`i<STnxYJ>QHgu
zvP0DV@vzxoA4F?}fR}3%euy1TiaZa)FJn6<xMtG4<U!2f#R*(G{*wT$GW4!n5#9bu
z2g_|9LhQgmtf=u2X57`pfh`U=s`R6n_75lXLyzO9Wyb8<ieR|vx(CnQvBJY&rjgcT
zPRwB68X`9&9{par6P?&&a3vrQb@WU?7&aRwwG3j*7On(ck1x1+Z35c#4rV?M`$1*+
zOw>N{j{Z0E7@x!W;)Bn!WYvorsMR!M-|~9}&68Th>OU)7({PG@3^rpaTD{n(bp<lT
zj|BgH`C{+=szm*x0clJc$dbYi3r-Js|Ea$eAJ6z8{7U9-PCI#~@Y9T0o`|5ux4MJ@
zFSo+#EAvTJvoqT|I~8p1dK1$@3TW@{jvcG>VeY&v+A7rG=P)C(`->&(-!z1NSUG~>
zp?LbQHi_2${z%nlSu@Gm&mq?%97fn5fz!JVQ?1(r*e2_FkoG*7h<+IhCdahc^^`;S
z=w26GQoM>GkxL<FH+Q?wP{eUb!B8Qwo}3MFV)njQz%*Bd#Fq^q=hj5h{;B&xapfKi
zJmQN^d!-4QZzGWrQdHu5bh&eTBh)><3r$<f&{8EGKiWC6*286DiLLMK-s${-ERQ_2
zs(b=PUEOdbq7vtXP~84Ni%jm~j%@BHkPi;VvxD>Do$4$+x}9S6((Tys?h14sTn8`K
zRYS3#4~qCq+r%;&Lp(;Zz8RcRy4f2(hlb!w&Rj{e7(<Sl9-<HUJo(?HSvYXGD_itb
zn$_NmfaJgvaH^sX3Ic4%?4h=7{&q`pW`q*coG6Le{lHw6V&IN`7-z#eL&TX3`pHqB
zyj755z2ceh-^@D@d>!amvs|G1R`7duG;RzTPL^c%i(1S=&?r%ku8m&_H35@INJ>0T
za<YJ*?VPDrNr+7599E_<g4s*Y1;<T{D&_HP$Jqz8c9#~v>}V0kQQ3GmWU%0W?;I|A
z%}H2SI>f~jy$R5Zs6@5ccq=uu*?f(vW%7)paswJ9Z-AHLAk=dFMsxFZf$Wl|Zky_G
z%^@dJ``8H@1~&<P5@)H4>2dro-4DOM`T<!T_PCDk(BBOhOy;@lp^ti-aluk~*3?~u
zCk{vu$pKQ}W#3M%@9!7pdPuNs(%dhK4YV27VDP6Zd>_7sRI9#5-vO3Tef%XZz1$1(
z8)9ha=c#1!ObSJs{`5>57x3sGMK$SUh&snPoGxQo*TC<f85)Lr=?wPmq(0cJlcwEX
zm+<W@8`88_4*b<EF``fp3LoBu?!9wZU+gW+sA4oNd?-A6brj9#X+d~TmV=YTY+_)h
z1nU)YFgfQ1_*G7ab(P)_UbF<W?y1A*qhF|J(0(}JXTcsy>O;KQ4f3ByG?=~*fV%Xr
zXzrc>BW>p6b&cy_A=3#3bJB$skM0Xyv&Cp&tj2!Xe83aq_h7(bN7mcnf(o+>Q9~ct
zh{qk^uxC1!kG}_3>Z;KB)@gW`Sq|hIcR=f?FnPWZ5}U_k?qg*VwWtI#w&@WmiEO(4
zdp=BP?!sZSIBUwb4eHKB3T}Bh;_SgkFu#NM%a^VvS=STkLNjZYufqM?@gL~afIf7$
z{fu$r)lgC07IoF*@PS1dxbzzc7j_I_UlLT<g=eM0?+L#!fpZ!9d_M{v?^dEUcj=v)
zQ3M@3-ht6&OR?l6Pxj)(d*Nl0BWro)KsIbJU`bW$=}}b$;{NayYNtQNtdf%`ap6T-
z9p@Wrw-^y`=MXHIeSjR?rp?`J3V4cpL7wWR!Hdo=Y@bojJ%~1}JZCgp`MM4zsT#EY
zN&uIteBSZ0B<&w>W5q-}jK060y2)D*y8}FX%y(>=`=lXT-x_wx+`(jbDJ%?c7J5&*
zp$$DngS86L|M?hj&fNn`mQ5y$7j%IR&-Kr0He)&`iBLPT3aj_YlZXmy@*+Q#SRU6T
z6@wd5#OEM^XG_F~RdTT~=qkCWJdEU4jV7p3B*bbS#es!+;O6d)^*8R&pv$48&ngZJ
zF1Mky^#IV_dH`<9M}pO5L;NGPnEXyyg&TLS!p6|g^uw0pG-fRKV(59`B8BDf%yk0h
zIYo(H@yvdM5G@GXR+05`g(#zvNp<!t1eNy#h(p>ZI4XXE>tp%nKt&n%{sneUzJxek
z8G*?^_fa8{&(^f9$jKd_vClS!#^1`sbB(36^KL!rhpCc}iwkgw=~-OxV*%^<yc(W<
z+0Ig{X0e><rMT<HFqU=igz)X20BaUKglBV$dGCA<9=dk~%MD(FBhOJUKCX+mt7Vy2
zkhb{i&}n3WzdqSjRfBzltx1W_D!es@vn0D@#a6ds@mL+7Ik?FRI?bFJ<FTC%D%cLE
zdjuHO4Y=z5QF`icIC;9{54yBH!XgVjbn+I+=j#vPY3o{2x?upHamukV&%Z$4GAWSY
zyYckWV_>^;JeJPnIlgb>!Scm9`uVmcneglkS@lwmRR<=JDg3KxbUI;e&4WnGrT6gk
zjUPN&9gM{^9L^*Z;+@2KY?Guwtfz;QC9*Bzpl`9jv(}>bDc@)*or;%-wL{&PKajpG
z5=U|-wL?6=!){uQDl?2(%RqiEZXU!kt{lPTTiW11Q95m%IfAvVuo2ue6~Xq+Lx`H1
zk49Tf*e8<;)F0%BcNJHO;w4tWZ=OBxS)2-^BuBE<afB&7l%=6IbtsG1@Y!Z1a#gMZ
zudSGagT|JiP5e}JOV?)mEQ_#xT(97@=s)2*|31my`a!s<-6guy!JURj_H*}Y09hG!
z9k<-pMU8g`uyUjvX=mzep2ttYVY?l(+mw&qIovb+K^^*ECZGu@V9Mho=$32%_e18e
zd<{wVM8k~bN6GQe@>X*B&2o(0{0@U=+oM-+1%35W9bCpQg2|d|NwnH+EVyO?7`mB!
zIG2D9?<ZhQ%^}>VdjZ2g#$%Vt2>L5zIc}&O&mt`Knc?n{B%^#IUN3Tkdy6|Ter_|B
zZMiAf@jD#Mb7m5HW6BEt3}ydaK8~LM2)x>!3l{!0G&$Rtd|RG_`ZNr~a$X6txpf%l
zv5;g(@?AvsCMdQY0g2r8Cg)TLThnVm>dA1-&)rO%`U~irC}mvqmcQ>W`r_}xc{oJz
zG0L(bq{k}+hq=GRtg4+5GRK%RQtpb*%jDuD=)uWJ9A~$<AIb*wplFgVd#tb>T{c%@
z|Ce)U{>&GWRz{#`{4ogh=CgOtarn9CH0RX}WV2=(!tY)E=oHEG2lwL~+I4@T^g$!i
z-#vs)^zlc%FW+g;v^YTxPe9SVBs5iegSz{p&}739bYEl1^sT-L<-EIGzCxK)e$gV8
za~?yqsU%*@y2_o&->~KV2kuM{#Li__<QI0(m-!>f=)%cDH-9!0MJXV$JjcOjy)RjE
zTA8ipey0g?r=YReoD{vYVDeI<;q$-yp!pybs-5D&g?s3VyhfAwou|p-2OfAKYamha
z8;K_-=;Dj}t@v^xzq9{-4L1|E3AVrX;Ag%kGplo85}`KbOMdvGy;CNPAG?$k9uncL
z^3&KfQGuON(1KO=XQ<JO`H){)M?*ZlNrJW|8cqF;I@K3Q@7SfRNHs;A+BA#ov~wh)
z(}CEx@(fsB|IPP@ErQP7g?J=71PX1gWB$Nz*!MRC6#W(fmlP0kJ`GQp-4w<d4I@2Y
zHA&XGyEwm=?+z-Zn95~6SUf3@e-`Z7yP@YWK(!7}_((Eu?G7-Muw!UHpN;;00L^3h
zJ^o!b7EG`JS-%Tp%4we6YYwM3UreXXw)=?u@+-7i@jOxmW!T2|%X%im*@B1(WWc~L
zFcT}WRev(R?fZo_Js&XOiXw(7_2F*EkJOucAC4syprg@dygpc#`36^^%ba4+QgXq+
z2ShAs_FMezDB$bHxiEU)ReZQWn*Gj<!OO8)f?7ceOz(V!a|5l()6Vyz`1E0HZni9Y
zEb|r`_e&79l&PROaTYE!0#f`el5CS-LM$g`2)6oS=pL>|g7&T_o7`oHkA)pv>dJ=n
z&`?q|z+N<M{6KQ<X^kl8vpPoEzoU+&9@wg=hlbT5&^f6=@F}Rm!m-+D$Vze4t21bG
zs25!POR#b9LuwnFNK!6au)GsCEPT&2ks?o?xc!<9DVz-wVDBu_ej&?LmG#NQo04SY
zBuTbCZ626dwQ)YzF$_JHB-H&Hfa0{3sBJZwiiWQvZiB5s*J2R48ks{|$Gm`nv)yp}
z-gZ6%>fxS|cxZYtf%)B5B#)jv;q%2IVx>;b*-FX8A&aG1!<YqZ1iu5E&^&^Br)r~R
zSqjPBnTq9`l7#Pb`_U+2B-V3Z>ypGbFxIUP&t2I_whxK}Z7&B5mf{_aoll_lp9(&8
zkSE4hwXwkLH`R|ki2w0SKuNt9tXBAoX$52P{y9QcZCyqujG2s=Cf)+A*LOI(uLe6L
zw!+2zmRR$mkb7<BlMuO!*tp0QJ$RQeNSnJxc;2dU+-sCrWLhTs>=fs4jYO&IN6Aft
z3(%Cc1y4**#<F#TNorjf>9$s+F~$Z=vmpe>54!``&#E!&jCgRKoJ*vA_lYiK#lZsZ
z@lbNpCBhEA*UhkH*cS~BgL}BY*9jx1jA72YQtZ;iO2Ori0U32m3bU+oF#GNs8f^3d
zULJjh5(n$>%nWJH)wC2%sapYoLrx03a^}z-5Q8Pd_u;{hGVGd!IV#@W#)eN{33&|_
z!sX?DWNuzq`LCa@B;8CFO(v?~qy7w3>vv?AD-%hhwFYi@vz{b*1*0s-4tE>mVe++)
z!gx#04RDUaxaBuE6I-5G8AOYx+U%uGvt`J;edU-UyhX2vYq3bThT7bjDwHhh!ds8`
zV!`8QbaH)@u%u=e?J^xHJ{vR`T$av*q<DK$;CBy;kFR1H>lK(!{S6`G&M?+cdb?ce
z+ZkxGT~AzgPAAz08F(?g1D!H&!t3G|xOVBRU@&bKQMnleQmfsG@n!{fX~20pZ_5zo
zA^nTKI`AE`LS|umu(j}$yc19PG>&w(ltIhi2k3M>i&)QS`2W6z^%aHUgO(bUO15Kl
zmu}*r`;YKK!#9!7fA<9)|6ed}o)xh$$j6?KE3tSwpQGe$BdX6o;84qSl%KOjxE?bX
z?K|IdPA9`pyq7cMoH~i2tq>jb1VUD5;U@P`^82hYQ)yp?=N{*Q$B}U?Vao<4v&4ot
zCLKiQbprYhjvzlDyV10>J4o1)A(*%L4@})Yl*IoTP2-Lz(1`De_+JS@*9Ypj<jyvl
zu~d~=WTeyX4@=0z_2<EnyRjO+-30Y-x}-r{OY~&paQ0yP6KssVf?kSu1RWCv9F(va
zq`sX;+jLWQ`b|Bw%D;y{oYic&+Lw&y9H1$Ct(n##8Me3}0+#*Qf@>d26G4?5d-vbL
z+au1xrIjYwG1LVgywW0F&Yfb@k@tj|ZsXy#cPHLkTnPRTW@0Aw0@r2<V!cF@xGA(_
zjY_!?9*gkcVj}V0z8F1F9yA&_Yft<NpYB~stX9N|Z=bh>9zOTV<7cuD1z&Jc;z%$V
z9zZrTFP`NEw!yrDHb$O=i8^LXew>!jx_l%KC*dU4?lx8DbI<uNZsO}L16Z1R0{R~}
z6s62tNgfV9iTCJKa@ng5qcr{_qiol~KfFr^jZ$Q5Wty<T@<n;qpqB!+yn_n29}tot
zf%&I@VeaDB81i}&SSKZ+<rF7a*7Xjhw;iS7COqril3eC|NtsG7H^+Nt4`JJ*J*a>9
z6y32xnyGFWNvPNux7^x6dZUlPT|Vn5zHG$ujO)aS+R|{^Z5=sZc@>8L$QFXX4TAOM
z%{bBh8IH->hq$H*ZRYjRZD05<!B~oAFOYTcd0dO>cBjFj!--{l3L(95X4rOgF3ppC
zK=o4eq3m-O7%HffirPc)H;13sAD+PHd(OgBi*2;1w^@|sI20satJ9WC#Tb}d2cw*0
zglC2^_#{&iic~IAt^R1?rOX0&5d93bBpzc$`zsvz<t^4|euJl`>G-$p9(GO(1K;E8
zh^?>{M?SxVie;@L<=#uEpB}&>|CNHZq$HDg-#gK{Ae7i{sK#9fv*GXkO!zoWk+`1y
zhS!QrAt5FSSLRJ-!QP8V!bCeT+NDb`WSGH~>BsQzk5arGEKPnaypQMfUSK|-6K}Sd
zN}f7@MW?<jVxY&lHSe{Vq2E9<G3_lbzaI{cQ8~QpW+7zCRifP6YgErS1@G~kl^)O3
zU%9Ny<}QpO=>a;dqUepd@Tq_=uZ?7uV^m>`raQjXkA|AT>xJ-`TER-KKpfDr7<Ksh
zs%Bb0wyCI--ic2@Z8OioKb9wfdZ#Jrn}QdwCFAk;8wHo28tm@tK_C^FME!T}6Kn5z
zNM|b@#K1{4aJYRWK_^L4yLK=$-%^fICtX-IXML<T+aMI3;yti`3uymED^yGSj7K$w
zqSDGP8d~Lo`ul%~o7zQe)yu0e_WE9utMLJXX&O9xolbsmdXU7c_hsvIOR>aOjG?NZ
zsrlj2IKS*BC>`2?TJ~4a{)U?Hvt&3VJWCMwhAXj%!|!Ozky6@Z*$eY$X|n*f1HEM1
z`Ap#r_&8Ld3BOk~j8!K2J*JSiNs%>krnR=;OVmwNWC_ow31TvuwMjg}&<_%%CUPnD
zzi<&ni{4<p{z@2cbR9MKj3d``YEf@q1Nv*Z(1d%R#qLq#VW`DLlx)6&{g*Ccl$$1K
zbFR?QU=Q$?>B9F@F9@G2KjPQ)+1z;{#f}trfWP(-a-E<3w^~M%`h=_0+dKvgjjWjG
zjqzmWrGc!;xg5S+2}b%P9DA+5V76)$*?xWlNWG3AnW(`wom$8?-4-x^`F&wnn>qVk
zVn~9f@>k|0f8mkIZK|{EELQWbwY^Y9XSg(hlTiS!QH|g%jcR&xmLc<9I*L7PUro9e
z7UIe7xezzljkq50Ve1ADC0h;NfW>-iY7##cy2|Cnb4F;f(CPa~QS)DjAH9gwm#dM^
zdn=&AK2Ox=oQ2)*KMLovhTsbBO9)Q9D4Nk_#)PCI?A?(A4{|o+FU@u=9Q;YV#Kn;`
zg~suoxeWR`FDK*kCZoHu14|tqMUDknk#UJJu;=td7RG0v6>Y;XRqBE0m@5_j-XFm}
zkGF>U2PLTUb32_RpN)n4*23snvWz~kC2NX@V86^&aE!Sve6Pv@(*<VCJtY>EDpUkn
zgFjHuvuLgZ)4^}EKK2hvBDuSIF|M|UhO9hApGVviyggPBtwINsju=G(APG_n=ds2P
zPr-L@4gP%j9X%%%<5mM37K+ZKZJh$s*NcGs6>D(gtr+U>r%3J=C1Yx#JJ!|>WQr>d
zz_6+e7c?!yA9mGf{3jgZz8J!ynf~O4RV4iS{s4_N)3B;$J67=W`u*Vq!;9s`2?d5s
zm>mg^tS7UPiv>uHX~5ZGB3QS*9;emk;(r6U%j_cO=<;{j&pB@3+dY{@n-=3{j}|!A
zuE0)DSc%nU5yElR0?duKf&CKtq<wZg?Yd%)apms9T&Z5HkN7Rvp6A)hW%e-L&ljc4
zZew>(EJnE?)ck6tLGu=pn5eNN+jN#F`mLLAeWC_iXT#9$V-lLoT}<MB$+PrJUm!@K
z5Um=@#fo?eE4BV$@5cGqIC-Y<_Mtnf-5x}}>fb}yy~9{P&wv>%xlZ(sTd^>`8T=kJ
z81+|{!|41DEbbmoGH-H6HUFCq7HX2tP3o+$U6~orT**0S@|YkUk8&T4SU|BH9vx>z
zmMgvjJv|8wEaV-p<3m}ZQxqorYN3BWCBaUURJ=3J7m7|7;=Pwk>BjWC*r;&~V=PY!
zH0vwuQM!gju>)9~l|C92K7iFdqj2>)&QI0Yf#<GH$8GUjp!ItsiJ!~oK|2lzmX-Bj
zto{=d%(xR&b<9NneR<*<y`kit`&ZPpb!4IItw`#@T$+Dfk2rlXWp#@8!0+2ova`aI
z=iUy3`NYHcMHVr*=n!3UE1m={pFm7T-N2{5Z)mb>A|_S+#VCs~vR&1l#dr;7oBx@x
zT>+^$FRvK~2YwcW9i_M}Mvu81^%3q2HzM}Kj|r*ets<X}3qp}q1j%OUV(0YJwCnUo
zI)Z61!|f|*g`>XkB=;lEwW$>OPiz(o4beF7dM%D9)h8SJQb~W#A9};uoaG*JL{Ak;
z=PAaL6xU?z>)JqrhHv6~tN<c=ZWQC$U1AV<U*tH%&LJxCDlwWAB?@0Nj`;Xq635(e
zCuv0qQ2Tc?@8>?me`evd{?|&JIyxQRrr8mt7(=`)9VlMbyo?m}gwacpa^QB;f$d&c
zg`Y|W6O$Uwj=ZJF+@te^<R=PDR(1^hn?4d8Ro~ESP0Hl>QqFOinU6DPxwABghNe}&
zFl6p;(C=OcE)h|1f2<tK?{g+you=q_fahy-K5~z@DNFn#!va4nCBJ3QP}iJh^wyio
z*7WRwq+AtDQ{F)|ij*-Vun`@H$}*edNoZ7=B6jvpC$A3Zp@xPCQ&mE6Nw_;0WJr;k
zL({N(d=V}>yazJ+YlWVSLB!v@U2OF3BH3$W!1@pC;;@WFE{Ch8*BoWorGroL{n_t0
zeD@xlbe``AW50^u4*5%^=Ip_Zd%i;S{6RvcQ6Dam7%oaUKbSS7b5`cpPdF{JU9{}o
zd;B@`9X33jPZ~T%k_(!jxv#xnXbT)FSh`6u8nQz8)+ot5g94#*%mO@WbsXKibD=kC
zBAFyB%WTu`qIfIkA;*kIBZoxTyD|s$W{pO=N{hWUizT1FXtH%JiP*E{H8s$WAP4MA
zMB3^<@bB1t<eA<>czvc8Wu6uaHOVG$bHPyr--&F9Sqv0&m-CF_yTRnQCd0b}n8_Ix
zX0E-S+|7vqsei-SGt1p@!#NlI&ZJQDcz+z(dLIAS6k~J)&!G8ChaRyT)z^<93pLb<
zVw*Kfxmk=ORm3p&(OQh}I)hnD8lWrZKS5%RJ6J`Va(6)wW({dYkL9=V!c{BQch3qu
zPuL1xA(}YDT@2qZ?W0AX@-Ze@o1|LZ66mRZ>h3umzSi+>nU*!E95rEk`+Ly&MFT9m
zvz?Sr+$@9^?Z-*|YRpnm^goKuJRGa93&ZA_BuSEp4247q?^!!TMJh=uO_E9_A(f=b
zlqpG)BuOP9$xo8Jdu^H}$&@K+lB8KGmH5v0x69?daL(Cht>?LK2KDO9P-_ZXRh9_L
zZfEgsRzK$WQV|-O&O*P#J@lTX%Zf;*&Hf^3o|ml8?mKz0uwO&iH)us{o?S-!>xS%4
zZVpKdawItlyTMBDD5PAEfnWE(;xEla$a$vDR<)O6(b7(OCnpx8zLj8+ca^})?<g1e
zZwWU?|2do;8cPZfH*xyo6G_^nG9mSfq&MdAjIDBg2sBzm!zAb1bbnSN3Oem5jz>`Y
zp&Kn6C&}j8E@GaJB5Y0e3oJ_AL5jYuqT*8gJZ6RiIhfvyQ)ACV(fC<5@gK*aq|a@L
zc<(}J{6#KSWD3cV5ak}9`HVZAjacXR3z*iq0GscuAv_HjWJDFX<C|<)<z9a-Xq6%^
z5iDnWrk+P_TOV@r!WwR}@kB=MaqRkX7t)t*j^&ossPb<ySgkDp*Lz3N@YiEVu!*8(
zo6{jwokGawMDk*W70Eg^muoe!gAVlws9ZX?ylIIQSvpmQxn)U{bGsFA*I!ALd2Pi!
zrKQN9F+6MK#zLXf>Ei-xauF9y=6&;5OSrwK-0_lRG*p+{kUbB^;oVJtFfiE&ZVlYU
z;8{~~>L^`u+Akf~-!sFXcF7o(vX(PjKa+XgX%SA8ii5Ak-JId(pXjsaEPkA-$kv#K
zk^Wo3B;dFx*}HE!c^aI6VSZhpbJCuyvOfq>*Zf#j{1F-=SPZ*&j%B|76PcTYIn#eG
zi=E>}u?^2U@%9}Nq8F^qwry1)WmJrXas^QHcO;j4ZwD!?6Jgo>Y#99OU?hLNM7UO?
zdc#PN<T*cYw=ISTmbqjZpPj3kdPShzHU)MmJfeG2d9T4?3ATEsA}BZ~q5MlF=HS|k
z2O9dgJc|ILeB&T-E*i#xLK&80e;oSd7=`AOAQ?ZAIE0KNuRBw@cMda|+{9+ye-;2@
zu9<-D)@<T?8+f@q2hXllVJ~GDl0`2@6A!PEENWsHe{Oz8t(YOBQesSd(k0F#{UnY#
z6$1(DVmQj)LFn(l5OIOO?{+oe%}xmtx|PsNXY=SK*)y1);7uOTNSdSd8SWd+XUnFn
zMxF8v^i<AHSZ1AzhMV*#`Ju(i+A0M}p+~4?SP^lJG!QJ>Y{aUT>GAnuNtSSO1ZIie
zB&V#WGqqzuB-YoFyI&eaKV<v|-UrI4S&R+&={y&sO(Y5T%7ml}qj8I;GV>YN%DwSV
z#%XEuSTJ8|a<8WJYw4l#{Qe)fdMKUyC#pp%h76$OpgWrsP!1{9s*Lf^lJp}g!q;!3
zsdv35S(zTrGevp6m%kI4_>d!YuE{iB)Q<gBtL7#R|3``pzjOOGe84Go{nX!7lWcou
z!4%hzK{<Yh{yq63PB!?3lcYcLUUbAJuLhJGm_ph{ALBxLrLkz9B>5^m3!0?FvFBM2
z-=}A|VSOT$@?5F%mkCg*(Zy$^M_}-V|KLk`8a|)Yho8rfVP`+-u)80MvD7sYl^E}@
zsqW=u{C%mmv^SaS#dEiBX_G3=%fdIUm6&wvk-*HzlWY2sDNMOtMkZw{vmav@!N^;T
zY^e$++GZAlExQIF>{})LxiXAWmNn2iX$~EDTPDP&a}a%M0yDQ<1+Qj)L@pwL93GfK
zTz`hyTyB<y8joRWxahL5e8LnoIeP)k93Bc5-HIo+>utE*#c?Rr;!B@AD}!~pBS}X>
z1*c=}OJr8mg4?fi*h8Lz)Ic~UTs}l24oIR(<U?5JFF};PWMO)07q^Ia$F^;1g+h6I
zG`~0&Y^zt%1B1)pK=N;lIOs==TRGv@_xzZ8{WWx)QH+mz#Ng=|L;O*3j`!7%#I*`$
zsJ|$k>Msw5uhWaD%85g`w9J%U@LG#=x0VSOoO{6q^Pkyr<S|bALoSLfb|>d~-c@O7
zI2wpfg&h}<bMz}=lh_*Eqn(TO?mRPQ$utrcJcF?jVyGvzn9SGS0*jVvvhMmy)K#j$
z(G_B3vAhoYJdUHfZ`YC3V11TyCk}f|%IT>$=OM*wDLLVo&;7J0K)wBKXzn<U&3-bM
z)XtYjU+Hh)K5YTHQy~k+a_>-(nSmmI|DW@gBqg_3!~Xgltn535XRQQyZiEslZxtn5
z)+7><duwP4iQ{bc^l(~>uTnpSOuTSFiIrsf(E}QMF4NBhP9%+DWlfEA+$TdQ_+rZ<
zE{sCer=PJ@>m6p+{{=a9MJ6&bf&`a`U@KE4o0Es&TGTAI)n_-_1<qzG#goCK$DAG5
zYQVC;%QF|B|IlovJ_`(yfy8!2w3JHWO2f0bk416(4$y)nzq*MzO7HRfwBPjo<U&F3
z!z?HbNyMkKilKCJ8?8I#0}?uXzwGl~+*4-4$^yMf?)??$W%#IE_23@Tvq+ZM)=Qv%
zT`}Dd5X9}il8EHTBkWzPMJ7BHu<Q@MbnIzG#@st0t}Gu4tFFSf2>zVDEety&4=}qV
z1s0!O0tbJ-!>nCVWXyLlqIkfTcnloJAb#)v`_E4J(Ki<QV)&h#_dG%JfnylbYC-S+
zSAuB=YH?-y1U$2g|GfQ~r0tLh85k7dPR@Ua`+_bCeAoIwx}P<PY3H+9lZ+w5L5`VO
z@8DXNz2;U~_~OUNMxcM+(p4MdP~?*|8($O6<$oH$u)ND$k;WJ{y7xK$p3Z;Ijvr6}
zGA#L%7WvxihzsX_$NcWi=v6#TAZPZ7bLY7czSES5{kivOF>W47GuQ@~d54m0{d;^g
zp$L`B-GqOR^K-|Sk;L=RQq0V|iywOqgS-C?Fg#X|c_L$3_O&YNFIESRmV7QSdp-6)
z+`}#BJ#D7v<#3*OCfaU(L))c)Vr(g&E1S~;iz@c>f0qO_JM2V<cNvk{XZDi!4KCzg
z!Ykb2;{>i{B24*OxnTLp1^9CHANnz2A#*Bj#lUw}oWl8LV1Kt0r~4K}+rt`9XDYG}
zxogSZcpEl&u0UvZ;2OQLBLg+sXMq%-ZKA((urN&pX7g-UiMVZW@>(g{lr1C!rh3#E
z+DO;eIYd%!6KI@!AQ<|u97|UA(!qvuE@68PRpR+;3Xhf9TBiiGui6An0VNn}FQg(u
zp5LXf%_;+1K_B0tnX4;CPYHlZ<802n<1IMIi?H)YUP1V_B9NV?#<olOLR9xlY|k!&
zGg*sqhp8cZX{-Z>SuL)6CCg@{$KaaT@ihD7FjPvX<7EA*Ol>-W9{Y>Jr9)$gXnr_e
zJJLqg`{Rh7e>;Rf4#6^?6Wr#Z45&-BfxihKxGSUcF?B;EXa0OO)clsDR@>jw{3U8^
z*O6o_a*w7W(?+t)W!JIt?iny1`pJ1KC&M!B8ki(+My5T|BA)Zdv6`ndv2{%_DsQX9
zr4E{?t9lHbmK)*U`tK-jpTuogYsXyMUkT@i6o9f}E+#d+vc8>Y$Kt+i$CYOqI6Izo
zKlM{8JoG$^6^EN3W9S84#tq=D`r~NTm&v`Bo=Xz{STT>|37iS<r;ZHyhXe2OIp_P~
z#J!^z;`Tc;^D#5Yde>+qewT5e?;LI&?E~Lt^K4AfW5TeXqG<A7h5vtlC!Bk8D^|Up
zM*Zau^BMkBvLM8jhL!fyp!{<*$9E21ujb$d|GtmUimBNFKa%lWp6%~E4*MQTvr4B9
z8y)$Ptnxw})(FH|(VwG2lS(<{{(YfFgJE##wgN18CP7s;8Vi^2&BV-?{9cdmC}<d!
z;O*H}=yFGsHa=HlzwGs4`ch@;`HAON4PEC-W&|={uP@liJDbfWB;q!!XZU=LE_;_(
z2=mp(vcZp`aAt)aio7xsinocg78^q{!)X=<UI-WF94Uk$(=lxC9p3Tw-I9IT9*<39
zm6$Hip`ONGf%dv?{Nr1XRdK&)hTIvRyL|y(i}Sg;f=29)`G?M5gSq)Rak%}<I3_it
z6YIUj*_-*t!Myh+crCg}cQ!PkaYY-aVVDc9MNz`))x(hSejy}#jAZ3+B=DfeBo+~T
z4flUaz%Q?LSV_JuTCQ7&A+F(&Z013VeyL&04RMzG<|rOW<{3b0GT3G`6Ou3UoYUej
z)J^#{_GmO?bM#a^CE5)k1Bc;e(kNm$KN*&eo=ZyA)R;`p6JeCh7Zi!r5H1V3!MiIS
zl&fx>N3L5=Bka>MCaQIqD70mOKA%}Iw``@at>-{m_y)3Lp)UKvXD@y37GO%jM(`QY
z2$C}@p=_Q6+goWx`X1lMwk7BB^Uzo5{;I_96wA29-->W?1@9!Z`^Vk4;&bb~Cu*cH
znpDkjg@P1U8l1bFB$i0AtXcY`^oKE%`_1?8yb@4E?-}o=>=O)0HRF$++H9xPeB#ln
z!?`@)Lyx_T#eYj<@RjUbHZJ}MMDf?Q0zcDrD_BibS|YKj(30~lK8YtyO|ZFo0V@dN
zIm;Vu@Lci?GAni;2G4jw%X|19P7;4l`;Y@gr=0|M)%3_Lk7ux7;w-s!Z6BG%`=lF>
zII%)DmYvplja`PGtZTy$!Bqo0u0&pfERGByi`6vY^DAdkEO>*bU8Nbd{30+koz5O>
zT*BK%J0RLImV9koju$u1AYFmlEW}HP9!Ok_>M79>t$Yjh?IMZc*2O#@{T+M=-^=u;
zj-rRp@@#{uP)^3%f<8P!ute?{ddI3`=Q1w{4?!?^*oqSu^ilI4$9UEvN0VZ0(J5^Y
z=$F06tc?o=<(G50-RFDI>Beor>#Kv<NCzNdVK7PLXT*wovN2mMfcSJZ(#Cr>{Mq&u
zJ)N@|&fnUGdQQqDdX5si*=0!#7t6En4gs)$?*b^mbQbE#pON47qqpox4E^l~;SDK7
zYl}CFew&UqJi~OR)NGcZ_8vU0=@Zxf9&X)YX`I=50bOszqgLWxg0_p9`ehTGkg%Ez
zX$Rumw~D0W&t&#sNgCgay8+|v=P~j2=~QgRTddY$P~wtCmRY>vo>)t>5e5#>u;2r9
zU7pEAKE8tFWnbva<xZ?;YcrZl=VIESa+rUhp1S0n6~5pzsD^LPV9#MkoTQS1UVJ}N
z#%DE}7{8;N)iUYCaA}rceU>w~+KQo#s|lOtM&w<@$eyTN{A#C9Zho1}L~c%E?{yOK
zc=s^go;nLQtA%6O?IxS-ooZx*>1g)gFP~4+lw!s8uc%_*AZShRfSFzA@nWA8?~7Uk
z=HFa#AXk@IiCS@8^#j;=BLmN0e91jo{E1T=F@k7F7?Ci?2!TTT3UrfwCUoqW#x~v_
zMO=1WMVY3@G%q9vr90nYdt^R*%1gy7>apnaxda?@<6!K%aV$wG3Vie>QQbL9c<oRu
z&CDU3%b**0>u$j8I!UsB^9!D9lFIE_8$k|oBJ9HJYLweB8q{Z6vu#P^!BbqGh&;H(
zrS4Q<5+52VTcS%=UX#GEN8-f5NP%VVuqH><Z6J2zO0HTv6)Odu!dq^&u>4aPIk5g1
zI6M}?>dm|0hC>E6{}+g=U7BqC<t13RFC85}WnsBRGuHMugK+T$obg)?lYT7bBzJk?
zw~Nwj#z%K>PA$TMtvBK2*;BA1b}3{Oo<pr|^FiBi1d+`(!S*~I3^_cG<^`AFNEd7J
zWc(j+(Krf;OBS)-)0eOzT@?P9{y{db8wC}!*|fXn(937e6$U8Sze%K1T4oXZDpBHf
zH5~7+cEh~ivLt*%2^?BrPb|k7u!1)hEK_VFddIl4>}6N+$$<l;xnK}?@Mrg;9b365
z>phs}xu0AtPRC}2Gnf%#L%vo`<7!GxXwf008CloSvOb;cQRN+3g)ZdVL0Qz4Frjs|
zpLwTg9Ewfiy=C)EK-<BBJiW<B9qo<ainu%z){4M=wIlTLI0Mqz(T^1yl%a0E4tZo6
zOuTkv!zYVK-YqbZQ_d14CWDjMl5xJIFjNahJ!rrmQ|98Sf7)DZ_$g!!Wf-LyAzaV1
ztDL<b;bxN#d};6puZ-PITzT(6RGJ)yoQ(vzKO^aSZA;Q{Y8Np*GZ_sBb1|$u9}C5t
zah6dt)J;!8@wD6UTHg^nx0ynX;173dWI0W%xsDFA71_JzI%sC+jFxeUShv_(P)a<6
zix#T0$5kVkoZU0%?7U5_UC*J}SyPh8cR*O*BHl6oN$CGtksUZYn{*59i6Ys9Wf7TN
z>e@N%{BBK5%CTa$i?2cN{BG(bt;o-s?qa;iY<BwOVLbLsj|KWUv4OY0QD$oqvG&|a
zvgFdCBI_>x&AEogY9{>La1F7qjv<>C^>NLsvM>-Q5x)_@HTeDqHj~33%pr_?3gdUu
zH3IBCD#E+$jL>NC6NnsDx%u#U0ts6o&(`Gd-iCegXk!_}W!(`M7D-6qj^smRg3nQC
zcH*-#JrUsXTN7i~G+=~YHKrApaW}_|gRuMs<=sj9$e)XCXt!R47;*vN6!d^jZ<|3=
z2GTKV_Y9Wt&xmz+N1<^?J(xXLWgC-MgKCcgUNq)+=x>{aVPhkN3!G-L?j-}9=k<kH
zt|87`gMOo_N<6%;I}1_QXQ4~13VY;}0>3m0Q9Csq!|rUfQ7w7S)ddgJHa@?Xq&=Gb
z5z`~9KcA;E4tvP)?k<?SX#^Si6^L!Bn#_6n7!sW>B+L5^@K4Msoa;0XONT3QfV#88
ziNA4F`eFQBcN!%creU$wA<P=}4Ehv!?({ro+@M^CNt308S6TvDP@M|6Ji5wvJ+;``
zr^m4{aSL%OTMR?KwOr0B8>|lW;BG$50PU1oVQ6a_)Cs1NwiZ_=5vs{NRl8Aa+kYUT
zC&T70L>irc1-fls!JjohsgmeX{P|=RC`E^W{ii(49m8|Aa->MgS$?Ki5iIzf;DsGy
zUvLGA#juR$3E0m!V`t_art_ZwN)*Zx>bREp_Fe|>ZAEyL-&@)|naEZi*CfT3$FW0Z
z7Tuj?P5P-E->pi93!d{qcKjLS{;uUb95vbL(*<0Rt0I#e5`&hDSMh=U6wr@(j9Uv=
z;qS{*q;{<mn{{t5@%J-e5p~f-<y!-K_;(4$-T4A$KU7Jwz#6l%f-&q>H_9qD<4{Hc
zM%}3w-g>wP{)~4g-eZzL{ee8$N5jxFHVhxMyO9S29r*8U3tE+=f$iPY^2J9BFtjHJ
zDsRbAZ&N2Gt*HwpbxqWGZ#g|Ux{PFfEfJ8e8vu5Sq%Aa(G*9rvxeha^l<soew`3nT
zV#^lN{qsC^_;nLcG!0{H=u7;#YXUBwa0fRZ`iX-BVN~JOT3F%7;F8N<`f_HMO@nv|
z?tFfU$Yje>*ML+(_pj+>;+WU?a8Dv6o!>w!7020Bo7lmjyT^#s7JffC-VySY`qA>5
zB^%;7L;c&+xDkUJ!9}H#OP+Cr>zi%C@{cWMf|sj6yD*ki+391{Knv%#7qId19$czv
z!_0UFlSltnu5Y;lh!}L2rzp)Q+52i}+x$#YQE><C!42DIOEVqiL?|*kjfbCElcK%P
zVOgO9Gwt3?K1@r47|REkyIvFerk;Xk{`X$?ly`Y+FJ^m4Ix2Ljv$_2@aGOFOdO{|t
z$_j&@s*>c<$!i#8UB_Kg09JLTN6_nZo!b)=gR^~Zfbx=^nDOg|@ZD5(mV2uZWZL*B
zf|ec&=662+egxgB61ct>Q_w)74&BFfP}_u2>}=jfG+f&SlOk8avbAThQn`h0RnfvB
z<*(4=^cm0O7h&cZ4_dWJggJeA5BK>lQN%``4=_iDbqs!ktrza`-p6YkzX|}Sk87w`
z&~XqbiLeP4b4anz2og5Jfx!*F^Jx_hm#r<Z*&rF`pFNA2(`A@Se>ia(F@c2_>9N0$
zN?~<!DF#=JWGfR-k~v#>m)@;eO#jGhRLqwr&Sx#yhfT(0!<;wVtg||pp>rH}9gl#l
zk7`)B#*@m$ZzbLQ`J1ahPQA{}<V3V`xGskYB=ws#iFdEVvQ8T|w%&{#x)Bdy_0wo-
ztuBcEbY|DD+EH@kA$1;kn0o$Kjqi13m`mzra`v7xQQ>))nG5W|G~ktxyE+@X`@+#`
z^JMaGsvkLRd=VF2>cX<>G$f}4q~qO7I##g{ukYo^k>D~+&mgFkX2n8IKICjPjaZc7
zBYJnwE|_`n9GCU)D*fv;k?j&4L6k1Oz$&fF0>_EE<ixo|a^On}*&Hwn&*kkP?icbg
zcSH=4Z#RVB2DiX)!9Oao6XDU#Kiob$d#3SfI+{sD3g=Ar$I5A|1j|01#${KP@J7o8
zjJH(6I+LlqV<;0R_PiEK`BuQ43u_?lgDm^()(8tTrZUyqt5~$K8~hB;VW3wE{b_Pt
z;GWD8|Bc=B<epRv$SQ&@<7RSd`&WZno;1Ig|H>6TtAhO7S3u*2HaoRXgnhX36Za(Z
z*_>MsvDNLX;MO*AQug*UCv)vN%saLcACI>n6aFwTs6T-t)(m6*4?DJQ+z!at5>Hm2
zHl~K1F~XhV7r8Mcn96PBxylbmvpqcxu)Zk(XXq)B5f@*<f^`0*w|cB~m!=3v1gfLX
z;}#S>dITgh($IwWaGf93&UZsc5g#uB1f=XD64#gE63<$UTw;x<B&L!3$0M-aFoc+^
zK7hqd=eTN_Mk?x%iVqyiaQ;OvbXb#ts##V{Jhz=^S1%Prmh)cqByqI75r=G`8iqUm
z{|<46Mtsi3$wDJ+`BjL+eo+|Lrc2T$jAuP|Iv8<-_q^F6oZn{1R`7gwk#Caam1k0f
zS{dq?;NFa8>7khAH4$ZGJuzNkCZ<|U<0`lEx!G13QZ`VBQ5~6hOmZZ+Iu=pQpYt&*
zUY!`K?ja&Ra@-2DbF|rF9P2r|lqTuaqvx7XzDIeOmK^=e1ztZyyxuuc>7raLKmL>c
zX-b5}cMn5Q>~n5P6+e@&yN+*<D6w~q{Em2S9^RcX7B4(BhRI{E)5c^06!j+K3x!Vp
zOm9P{tbK%|`|OxgY#P>_O9tiBBiV?)Q!t+Qk;pu@Voh_Oz%tuN@}_7cQ<IY-Zac@5
zq#zmg``$>hK2VL^wfF$}Mvw4MO&o6i*#Tc8%t*1;D^5171BsXms$Z%`yDtOi5H<*{
zfBCMF;56LVHz3A*rgfB3JHECHg3`9l&|{P;Q10Y?DTm82!^i<upAFE>=PKx~MMczq
zy9pVx8^n9n3e4L%ip$XG#IW=VXx`$0c{yFU??xmy)uItyl8ni$rXMKlpuwiEJVeI@
zXVFB{K@2N8ep7k%OriP%Z*q*k@0luFkpo7D(d>K_DZVHJEm_U*rJsMFXw`uKd!c6k
zyg<i!t7Ch`N-WwwfqLhT#PpDRu;_^qQ{lS>ufxxxxpfgH#>Y^lEnlF&=@Cv!OhD7F
z7Mxe2%Urj%((^K9bnMoNEMt)r@!4-l9xRh$<xiBE{F`O$^9NDJ_o{_`$7Z9g^FzUT
zmywK{5RC8n{MfDKZ$NiuJyqzphUGdhIGN*X+4kOOeE+MQJMeM{>#}<BS$H_gq`p9<
zWO4FqejH5{m13QB98E6~K;XL9$oqHjYC;uyeSF2mNE$-I%6rt+X;8Rv^AvIy^0A6%
z<QK`rl8f`MKyc19?#@&bPTAoer~IlKmTs)Yxa1Ig=p)arRh-}ggI?ff#SFpd@m<in
zco3U)RIy{PCih|EXV4rk&(fUtk&oItp{{rhT77#Cd(56;>jpDY)bW!`9F!!l<MLqr
zkRghc=9KRUIf^HjeuQ0Cj_k97AHL_^kinAi)bj~L-<OL}^&xOtmApT-V>_n6BTi$^
zDl&adDX02sHZk+HW!7V_2qhN0<E*U3xscRmI5cw`VQXw5yz3k&7=0J|b&Hd0;(IV_
z-zJ;;yL+*z|1kI9UpB<^&;8n)8oc{R5lctC<=o_WzPxuh2^2lS^+p$y*@|)`W=AHr
z+djdq`twlTeLKBee;h^!ts<&3Q^@_n3~HYx%_gPB5c{t0Sj}_bKH90mpAr!=szrkR
zaX5hyVtVXZyfn{Lc!MpUywR5N-XC{WZjR$rwqRB!sS7YA$zJzh+loi%EguW}<!eCM
z=^h*#Z_aY=q;Rf=rqHJ{18dktItmleb%nj~9eDz$f@ZOf|Kf1ggSn`k`%j=9r^jrj
ze&_CaIYI!RJG|4q2v=V?#w|E@2S3;z6-Hhd2QeilAn=+JGZ){?gpU_+*?9}W_~j&A
zM9cB0<pL%$YAa``coQ64|8gZqK0?acaxl$yV_u?WM8>d*3yYE!C@tTDVeR|4opVYB
zH^NG&(~N)cvcd+XmZ^caH{i(;VsLf$eUup(7A~1|7f*Y+f&IuRq?y&Yb^lD}8XZKh
zxjC{e%@VlmcOmCL_b?f5zDu3={zjd~VxZ15X~(#;sQmo8pzfEfpl4(c&-LGchI>Dw
zc%(Qx%y-bkL?@E5zxeKn^;B4s9ZG6VW3Wi}2q&U)om)2F3S1n@v1xfV?%35r`<H%#
zj4}3Py;=+uJdz<MX|8a?M2P$Ms4_3LXmFnIOmYRg;Mv`OC}H_qP<vr7nK~{KHn)vu
z{nIir`(!`on~}^-wD=BcSM^|;tRwz3l3+K_odQXIAFyFj9#(wH#OnLrWW2KqNoxH@
zb&v2_QQKL}IH(cqeq6%)CoWKz#(q?ImjcDFrC9ZEP3U45==R0;Q9DPK<{78+ZgvZ1
zQFR|gytKI0cKsMppp655eS%86i}X#-dOWo#5*>J6MwWX9lmzkIfM3Pj0mTJG^zAA#
z_(_(1JQje;F9yJ0b_ciE?mBSmtY~@c2hO5n6h6^uzzwH%l0AX$#QlCVXSqoaCwFR-
zars_&W2QLUcPEe*FJw48<_{M#*MJ2!6bn>)jfuIk1o_meNB*t<3@xULY%qQbR4Mz>
zYA-pm<=}O!`8ky+JZHF;_W`wR_aZTghU9!e5GhN!P(JJV0Qk}@s(JJmW-A<_(K#c@
zz(WMGD+vp8{?VfEhpB!=1@?~Hj{{YXpzk{sMnM=cuU(Go119i!zZ5VuS;cb(UU8DV
ztJeI%39_{G9q06<2WxlTrF+(E;EzkMz>Mde)hMar$pZ|Z@ZA&lUr(`ktS<ENY^iwF
zV<czCD3<MgjVpWMi8|k`*@ETHRDESMs`N49dCMOn|5ITpOE!=~{v6INSc1o=>Jf?2
z;;igSnJ}~V0<^6=j7IHIU>>v|XD`&iuGU1f&)J8PJg3Mi<rP@pw8NB(GvNN-=|HW;
zNzBC>yfw<6Y*8N%fwg^D6E%#MZeiHvJ&M)YM++mXBw0hm5Z<T{hnL#D7}qxyZyei?
zIsZh^Wm^u4Jgv3<`P_}<58ILCy(3`u(ueqDgE5P4)+Lwj72ty88}RJiSs0~s0+aq6
z!h>gJ*^Npoo+UAkGbj*6EkPQ@7MWoS6=iCM(y+EGh#Z-&MLw!cVDtT^vb>QltX<ZL
zd6;+M;^Dj0ce#+(9)E|EH%5_}u5S1v%oF}QJ_l<m$H9^7SGe4j?({&%SZGg<DsPbd
zL1iKq(441HI8)h}9ML|6il>)g>5J3&iX4H}O4Fgs>kQ^U&!aCtUj&Je2Lk7r(;<1}
zZ(Kf7o!$3Yh_lDPLcM$WbnBdlaOYG34%}bKTz=^w%T=W3eO1_X^(n;q;{cryq=@)t
z1r8{bQBBq3G*R&vW*&6FquZ}w{_%sP-F=v=+5MSb;dcf5N_G$j%eOF|e;)h*TUKei
zTQFq#gU))zx0!Y)Lm}UH`<Gu$-({6!Pm>qEo8iPXTTjuX^i$j}N6K}lND?icZJg6I
z4EKwV;5E(fa97(3rIWUh-FGIlXLI&(@%QD3+1)_&yBvk(owbmzw2p`+-QXGKJLsI2
z>8yF*8Cs+$1cz_OQRIRWt9x=uFiyo8UYs^!l~0|Jv+3m|)Xbn>+6dJ8UO@HWHuzB-
z2U35uNt3iL^Gg^FijLw`Ch{NJFXFjZLJ9Ko(MaOFU4-2`tqg8IBv`_!O;pD5A<myu
zg!`sQkik{_?#%ZY&fT4aCb}+Y=o10|?gV1${VXCbDM3`^b`a%SLVW@c;(di2(x-L^
z_PKl!Ec<Fir&jYkg2Br;T*14(9{v(6n`KOO_L`9B9e(I!U_p(_!@=lbG^qU4ffZ)8
z*!X@FW>+1C_dNk<ky;4tS3AMOtqT*>PIDKhZzmq!6WOH^^DyMtIlR7JmT6C4L-f_>
z<Nkz5I2;zmXKYl#OL#<38Z?$@JK76BUa)88i6hD70R<wJyaw#oMN#7|C$VM6f7q6y
z!gE6e@U7DVa8?vtnsT15<h{U__WjhJ&q$8)dI_1RK`h0mLeiT9xbv9~bN#b|YCn?~
z4#zHHTVMRa^Sw*i>3ffOe@+w(Pq4(_CL;Wq;T*SL;v_ER@7EoT*SK0A2mCvw8+WwX
z!Bu-h`ZOs9yhh2BPJtHFn8nZ6f_d-BHwl&td8n)Ti9grdLa_fRR-;gWrFX}pd#Wkb
zf0PCPc1~u62{%y3p&2Hh84dEq_M~b0GA6ddjIfj1oJ*7$R#xoAq&NxTq<Wq!<P}b?
zuj#<917|Sr_*u*^%th5X&hT<~t>E2zFC4J?j4wkc2~}H7nMmewn^S^g*l_4CWJ_E{
zgU0F1ez^uwd27lNm&|9{?pK5vCX(FUa3|gwmW|O0`B)j+z<mzPrDjq=(Dk4f?^oQ1
zCIg-&y!0@Nm&&m)RXahDk3Hd!MQk8y4VXPy%r<|E#12^*a+)Z>*lj)}`u%s<ux=85
z8n3}@w~j-DpeK0Z#AncNctiu$PjI=O(p;9>Try|JJf@)UDv(L=AxA3Jc+O%mY?eB}
ziS$k@pAc$93i#Y==_)g}j_(TfJ_v)og=aBhVmcIjzakiSH4OD7PH-;#&sC0D#MU@@
zVA(ATQfBVSNgOdDwo|%<fzQul?om79Hzu5h*(?;)jMk(ImyXhwgbTD>{TSV{)}3d(
zO^20d^x2@{F(UkP3q%JFaiKb=VUF(&uwRsos+%-O#^MjCld=}39~=SNHQmDA>0i<G
zSpw)UKvege2@*}x?Ca?{LR*jhg4THnWc(8wRR3*+Y=Jj!HW*EktNwByrN0S2uBf%~
z8Iobg;uP8K@Fh&|dkk%PwTKnfUFOQOtl+zi2)dTcqbK~-NKk<g`$uY#>^=JQqbQ$e
z9hFCWH~Nx#-vTV^kmaIJx?<IFaeRI25q>o-L5a&VAbt91RwNOD0eMGA?xkjIn4L?D
z8)I<$xn789_a%J^ecXL#zN>rNUvS?=iWvu(vOW6zvwBJk`i*%Z@J;LFUO%2eDp#}%
ziq4h_9=Qd;l8$Z~&}_vDblwZn`u0(g%V+3xqZm98+JHrWHCR%E7E_mxBcajt_(wE`
z_O=T_yQ>%i+vG7?dJ$c@X9nB3*$AU<2Vq3?Z(1UK3$j|O!04GaQ(lvRDl%%=Z#|M$
zok>T_)zZXsvo~|HawqvqkHB8xWHNMXC+hH-fYTyvsCQG7$SmrHA0kGiX;&7feXdrJ
zS;0TILziN=yfa=g)8Nl#HA3Y}JA}#Z7A&Vh3O0m(f^U3ZQ~j2Vu%O?BVpu**64^k~
zF7W%@IfyA|j}hC6&oMb^E>WH~7CwG_306G^$p(il<aG%_kqI*doi&f~wLf8d0+-_S
zRf*KwOoE-rGe?gX)!f)e8$j*oYdrsUEU_jdSpRcvsCgmJ;>Lf#wn;}(UFJDfnp!be
zmuO+=Odb9_{2nJJPb77A`m8T@JE)k8GNo`Kq_1~Cjja@GqAx&jU_X95p-S>Mg_DBl
z!{~Z^4cE_i_wKpy8SSW(XnimUye4T2sv33??FvQmuu7gC-ZYjtB}<UECZ<@q;UKCr
zzFT@O3vXv!MN_wJ<izY6d@cPCcGgr1C(fRMGPH~{9J-4u2W7b*-KXHb#%$s-#Lo?~
zx1q#=P_nQ3G~Sa;!_b4)Ol#%<7d&bOUA9e$=f2M;+HZi9Yi>gCw{JjZ%~~82BFbzx
zj;B7H4*K27CNIUG!i&fwH0#|l6#Ku2XJRrpJ9r`!v-c%hlm38D&vwjOxEjvSe}p1C
zpYXY3ab~gi4ZV}Lis)x6G23ik42w#<IZtIa@B3_kYxyde9dH{;M?S!>pC$<_JL<sd
z)GDeb>J0uox5cI_11}D};oU@!p!CQc2tj3{pzcqW_V3~uN0PX>p#V7Ex4z(_FN)MQ
zp|-X*iR>8yk$Ekg@^Ghc^-)F#KPj-SPes@h`30!{WE6AqAH(^pk7SJ}CPT%X<=nD>
zN7z}EjsD_VF!PrgnHVJG8h>}vuE<QTg5M)qM&#qorf9fJW)jhcH~7ilf~0A?VeF?=
zPDjXpO>+}?7D_?R*f28fSUAS=nKj?R7_9M#qQwWY$Q9KESb9Vrh87qQgDb~SMr#!}
z-#Z(^!t6-mdjz$Lr9{ksJF(4^1nri|bnx<S2rrw#{OY!mZ>=JzVV6w{W=|w#CTqx|
zOGhx^;4Sogm5X_bkA&}<KB2RV6Ba-Dhn?wiY(mU6^0K2|aB}`>az|2uiCm6@mgM)~
zc6=RioMXYBO2m-~cE6$Q9p$!{%i@vOm*~xr^|bS}Da(F#55{&?a5MO<wm*OUYDcDm
z&?*A!lSMH|(-W%KIFkNQ5i)y2Kh<n`jr-;wq2;B?XeGUqx=&h1^!RgA_x2pLRdW=c
z|9J)zYo2kXCoRz`+>&LESU^NJL=s=A^_*w(5gfU1HSyqGX$tKv_}0n@>v0aaPB<-`
zE0s-0dgQ{~54K!u`(h%yrw2YeRN)pG4bUq>Os**7K5kTFX~BxDK}(nIG#N*O236tj
zuqTnMzC}HcAAmvG7#OZo=lMn9T&|Wpiug{(^-h<-k9Sd<42@-HJw@1bpSzHj%`wl)
zOQ`fi3e6QSliKH!j62#R)Ux6C2+22aTC6keSI?tkp6Ie5<y_RzZ3Cy&64I)Fi@J=h
z5sK~FPO``=)K^>!3!lHj!gq5?4|&3Q?&7`jyWE63m$%Zey?#QX9f2K_MZwKn12%?g
zp!)6;0?meM4v$^|&8xLsTMy5d7hR6$LZk6`?q<w5C57cvk3yZxH&mGFNW5B4g8icb
z7$3L{R?Z%Uevt*lOVV9<8%DEq*(kiJ@)1Q$pVFENb?&ooJZc}hj{|!L%X5$OooLgY
z<bBOt2sm>aZC|Lexxd8Nng~(eHNKEU`5LfV^>7^6B1ugp_?&m~RH43Y0#30$5B5vO
zGUe<(j4e4sMcPKAciv|R+{Lr(OB{e!^X~R<$MExJ2O`^Y1p`v<VN>i0niVC%_Xgcb
z?3FRJ>|z#K$@B0^-d>_X?`;^T7*B$f70Ix~WD@*Tmin0`K~rTM{;W8J8$ReVt1mN{
zE6?I(qZns9KM@CIPQh1qSt`A%5+`1;VU}xVGvRhi-Y@eB&1wl*rm+=v4k)m@JI~QC
zQ_N97rhwYKJM_(tL=t+7Lyz65V9;cV4uz}9&g?T7cA%kLp)LznJ-&sR8tx=p;i<ql
z=@qI?>!OM4A~7p*A<h_;Lf!sd;PVR)sa$@sjppNp^ls)m{Ad*~(36ki#Bv^>t%S2c
z^7cZq;F~7ePPTv|$r5Oqz;hOKUcu<D$GGo$7!0nMfGz(;(1uqNFe)zw8|F<RgNYN^
zuYJbQum2Y34SnU@mM4?_zs9gM|8&kf=>e=b77D77r$|QWMC@NQj;2oDhaxU<-0@k5
zP=R+yWvx3e9C|j)73z3l#M4NWe3*@`H?$yWN``RrZ5hrvI~7A}@<2vklWSO32xs-O
zP(J1$Cci!h^R8Kv@gH@W#Yluw$u`=$GLf{u%7ll7!)Ug01ANs>6mEO56kE$R=-gs;
z*1u>Hn~_olHs_6)aYH<}Cz9uBD`#-4k|`>LyoXl){l-;oG%;@J;8a$bGjmT}))QyN
zl>J%-wUhXB6K9Q15BG8}zde9v6*cy<(!V^DpJ4!m;gQqv7`Vp~V$D71{?1c?!5QE)
zW;xp!s>>|5N8_KzVq|vdUA|M<EqL0Sjq?lZuxGz2?bJ@hc<UH6HW)yqPm3{L+Y`r2
zI<hti1spUV$qIgZqt&GZ&Y&>?o@{%IVQn`B3Q|{b^ba6a`=+p|vUi|HVH=B9y9!1t
z4q;g8RpA!VX)Lc%2meXwGLzjZ%)6r)T#^i#gEAollZSEI?H;^Zw3W;imt<997m+XB
z;j&A&={jj;R>rv#&Hf>j`6|zu(jA!X)5v+vjuHfl{lW!dU$~dMH=${ZGP5!}jha&0
zz(q{LjVS_*y`2k+t0%(r-!mEMDMy_OSrWD*gf3XF3I_v@q4wGi!MYb?nbSKzoXgLI
zYZG{$+SM{T{EMRHiU@F?=8Vb>|KNA@N)~b32qGt~VG)0$$d#v(5OtJy^`tAZxTA8U
zS9J}`@5!K#%Svd;v#;p)v=MG~dlP%!pO^LaHWu6qKt1|S7{B%aE^AW(95;vLJQ(0-
z|2`<*bQP?FJJDn~0uSjav)=eu(Aqtfl)8;(@djrgtY5#J>7`@stUnO@{U<2hcuMmv
z)gdTM1VO@y>DKi_fQbxSyL2rP>mE-6T3m5vb}j6b)gU3-{-}7!fUQ1ghyB5S>4vL9
z!QBsz^zQu{E=SP=Zr3knS-z2UacctcR~7+<@{5>Q6@#N5O(DA)+wpW~3Ra$WCgf-_
zcKe3n(4y^Fkktf_($;WwfmS$gwmO~M+d*sAxFNez2-|8;VEcImrg6RvFBnf}7iG3!
z&_4c*-~JFRlkTFl{7H06Fl8>MQo$rB2J=0FKv}W~6jq4CfS5CV>~RiTbQeP2`fi;4
zQJk&NK7;O!mbem(f!C+wxAY8%)@+3Kr*Zf;domXL`=C-EpS^fofG@2=1?ny1>AA{7
zxO21)u27I5=4E%`!NpT>zJ5CMiR3-!9~-&C<HNiYzzrwwy^bm$Z{dWd_jJ>ze;Dr+
zh1wsMqt#a#DE%7_Qp5RpY-cZ*CaFlLb*yK0n~Yevtuu%!?!-+GH(^Pt4jEh=2{xCa
z&~=?H2ClOqf&3n9w_+Q_q?kcl&n3+9dL<ZDQ;bE=GWm|S3VY$V1oj9I3qRMK#F&U6
zvUYkP>A3hCAI$lJ4~Mk)j2XXsQV8O5>czp}NitP0+{Z1Q(uQBH#}JDZ5yUDa95K(1
z<@5XkneE27A!Q31dNbIVcNQ<L6J?81_^iLwP0qw|21_<rPwEGFf5Xv8QtFeAC)2!G
zSDGP;wV30;a0;ki?!fcUccJ6L^$Zr?#V_Y~lYMd;+(#WXuIr2v3zCY$uihmPqQ`UY
zRU}EUr4m`aLJiGM#p3LX5={4D5*R5(gGWa*Mm^5J4SOBolWrE)>uo?01@2~<P#H=*
zw~>%{CTwSaKU7W@!}!tmpsngdHn-K4mjPjad7sKm_fH@ZI-bo@3nR7s^{-MrjP1Nv
zdD?peBK6ptq@S4x1us<CV6!G0@gx8m1LlE8>tu+W9u2ATex%#U4HcgV*xJqJVB&X?
z$6ncjc5D+gnkPY!(>Ba1R-`}dUV{(srP98&3-WSg*iR34_Gj!0(ypD1RVxiiT;y8J
zoA(0x5AFb2Z!LBV6_}+ifl<#^q0ed~)~YE>CcQk2`bH0Mf2k1+=_u2X6&c(Gg*I3}
z%NjEhTH(OFHkeYnnq~1LU#$~%#ID1Rw7S$|#^_93r~4GQ2<tIsP6T+~;yd8(dgx(l
zNiD<m*x<!ra^d|3RLh^tT5?|E&MQjT`*kLR6(d>XE#3<-Z8NE?X~yh*4n&)Or$`(>
zj~M%EQMtKYv_^k2Y*<6US!WN?s4xc`sT(lR_Xf;P{-Vp?RzlzKEbzIui0F!?fKII#
z$jy#}TIJ>J@;h~AZ=6Z1LWgZ$?w9}-8$Zz`(I{xuUP<~6nS!>vDhr;aOPaTL!`IM-
zM5_4|)~U=vk=v7K`Gx>=HI>D!{!Wn6bPKw!bwF%}INP%GBu3EP*nj>8D&96GQ}m~?
zF4JS&N4YOtw&+j%`s5pGS06_IL>qWGzZAW7eaOo#-gNe5Z&b2R1(lc_(3_fpQm5x(
z{tHJge$_W@6c`h=??x;|o#!71Mo?d7f$imn#6GS9jDDTK&t>75^?V_xapNj13c3vk
z_}zBVxMFnfjU-Z%qd9F0vGNZQBZ)RwDo6)4=$@#@S;$;~XFNau)0_$H&?=6t_>_%O
z)0;4U<3$vmJHVynn~}!AOh~-zz#>e>LiMV1gx6-l?~V=lSamOm4Sc|c4p*2RWr?$V
z4bc4Gbhfl|Dw7#jAz>GNiNmi7)T?gg=ZYTqbta$r>){L6OCDjCAMXaqP7suxev6yG
zPR5WBb9ir}z=p&$aO1E4kbLq7pNMQ6cSQk$4fm7#X<4|rP8U1Hx8loT4W@nb9+#Y6
zL7RP~$#U;7Xg_+BZvAo<oo^enEYaO~Ew2g|T@b_E4?N?2U^%lXlE!R)PnS@%gj>5g
z5YBD-50dv@!KR%Ctjf)wJDAXjR(YnJ0yII?Qc1RIx)wVb;6t2@qQHGjAJ(c$vE@5X
zz*5|b3rae$ezZIr5%K}GzE$#PmmU<=EQcih)tuZk1w45on|Bm!;ua@<0QH18ES&j~
zOVSXd#|{~@;fHZB=KW2aGJ8H365~hq?YKZ3jz*F6rH`qtbt5{By9*8Vf1p-LmUBwK
z11gR{de|&7EGrFN$=3MtK{@;jNkiI~4u8Jb<A<To5K;UCHa$GRb@=7M2;V78YmOBu
zQP5=KwGogpVgTYK^q7mYH)j8PiFS_Pp<^A-q6%G1lGplR_MF+oCh{px*r~``mv+$(
zpJLe9D2+8TvTWwXOGKmjH=@^RtiP?toYLJ8&P1VdkPJW3_P`h6`XFIx%K8Ep5X0@2
z+?Dr}nEQTbvb(<sZLMbrQeLFs>vgVF_6I);ZI9;0q-{rys9%sb_ZTj_t;@PIA7km!
zqvZKKd-nAXNBUdJs6s_E?Df+l*W$jyfe35f$0Npmn*4(5sS>RAd>4*gI*jFJ+aTIt
zJiW805B7+LvRC#;aGsJeTm5e|`Sl`+hT1O{>g41gn=}Fo<~4w&^J}imTLPQia!}mw
z6zK9E%H4*VM7!FL^nd#$yc?c`^=HzFgy<$teez~1k;AcbH~s_d{a;YI|1TtNdw|Xf
zR!r(v9llHBAS}DurX`{ki}uVHT>BG&pA=MJ<G^_EF_nWqFGjF@@$*!q*vCd<Zw$`I
z9k~BW7v>)u$7~JEaqam1Jj=D8y5I8xRgric#Lu*8pp7Q{%Ht$8M=|}2rQD-kZ$ZQ1
zI%G|p!6gkJ!tDBbD*9gwXxwQ*xm0oTala;4Sn&}<8swSdS9Nyn&<XTEBm#puvGnXN
z0s4EjV^xq8>%M8t7H!+c=F2DG)kk7f!%>#}vHy>A+%k{6>R*Jp4*SUZyYb}I3n^G-
zUkG6m+RW|8FyNJGtlne+4LR>moS%nb&&p8pnRlM2ef$KU%Tvf%6?2x-_ye<N2T-%n
z1Ei_&A;!y{Kx_ZkSm<F#_FHgp!o7s=ZK$$AOC6TZzpo^8=rFzG67Z}kmn$f$g7VWR
z8S>6fmvJ9(RZt)dY-;72HeSQ(t3D(n;~tjf?iWrRGA674y`YKTPT}&POw`?&2>Pjf
z51fDBxN@WsH0`qa{x$En)OBM+X41@M{v}*wz~_h7t;Ev3%OPrhC>L54ge_U^aBRax
zk~U(Pa!-H3;;_j?r_qcB`&{FkRJ!PG!**<VE6*(V>)>G4Q=B_nl#-q5(5qQM8|)cp
z(JjxcmUJV1k_6v%MsxOe%!pd}V>CT90!3t&SeMl4lfFe^{PR%__gPm`$wr<fvxLuI
zhFGu_GnJWh^JM07J_WO4%|YheFA$m%VpBAUWDQRNv!{;iVAn_dyYCGgeP;;Dmu8SD
zULwSH2mef)#OHmE6mkl(9z17d8VNkzOMf4fX6h>bG_9S_$3{QLiAjg?7N4zj)_u(1
z&kn)nBQjj4dl+f=QG|&{&k?i07W5wwg37WO>SD2tyflcxf4Nn_Y23o{3kK*cJVrci
z9MDSEh*?;Tf!Etop{PAdxO{ghZcee{5@QyC{_&?UWlASZJnO}Z&fcd}){g<FXjfs=
zjiXo``U*U5KfrPc{yE{74XO+|zlAY^vP%)@xjhZDv;qWSd`HUNDux`AnT|0M9wc{a
zg}_6$n``)wXAtd<Lch^7Sk}%etdy5!2Sh~KeQ^o0qwX?_>6^0l_m+5AP>TL#{W$CE
z1&EC;$9I=UGr7Pz?EW_%l7CvUU2h8E!-PiOS$Gu-+!kW&qrbw^KT%xc<ywq!xP_Ob
zwPB+!uzn}rRo|P#-RV;yw!3u&5tI49(||th9d!i1+*uCdwl6VNd>QR}cN4N*+c@P)
zex9+qiQ9YZDQ=3FAtR5agZqYOoZsIp+H5UJ`l}~#Z<e@#{@Bs@*}e~lepPWA$w?6R
z_bH5A5YBD;JDFYH5Q>2XqnYybd~V%Beoq!vj6HibsMTBv6!Yl?gKdeB;XajiSlq;<
zC6`d1_dxF3EntVN$APYM4V1L=fw%D)Tt<KuW6u>yA>SqDH$sq@8U>CArV^zkXFxIc
zB$v9zAMK)7K<0iWlAC&kYjD*jR}Ox~uRikypR*Ql2FcFo7Osk7@20Zb6}8y8HUn?F
z1qohh30U~-5<LI)3#c_Nho02h|2ssPi~SI7UbKQ}b{eC8MmugOoP?{E?m=bGse%W5
z&UBvx@7<o|O1?^};=Ytm=o>JFweDU?^Un{XE0{8Mqp`5uNed#=7P6G=m)yN~$Km6a
zU6c#y#>xje0_|X)^|tV)K%^nIoJq~acJ>yvmG%gipOIm)BKG_aZ#ua;ZUp=E(Sz(C
zdjSlG3!urr2eTrY_$&($!+Cr@)^!<+z0YUgV-J&}p&x>?lVphV-58=7a)q0G`Zt)b
ze1lgclv&|UIqKSM4u7uQgt?x(u;1wmS7g4K8cR=S%lT~A>HHCBo;HP*W@$0|QGulM
z+z~-ck1l8);*eaJk8K@tEQj|66`XXzd6Dbb=vzzBYs*-2ocE|%3LbFDn*YJOb_af!
z{R)3Qzlhl))6qpQfLJS?g0K{6y!Stf&cu<buM5Mb5Rwo=k|a}-6uD>Zlp!HWLXt!@
zmx_`k$($r9k`OA2l0><CZ4xS}RFV{l29^3%RMPaF??1r3&e?md_kEt7V}<f*$y}Z3
zX=vG*XJuTLf|)tTF;G>4Sv6<Udm2(iD(gG+@!2-BJ-xVchXoPr<m)l=+nHjAGy9$q
z!OuqQAYz;mt^KV5FLE>S!V^Uj*1wV^J@;ZVe)c4FtR+iPV(`gC54sDNW7U^p*l^XK
zDDyMMlcg%8WAlGtGG-rf^;m?vCU3+a?lR=^E-kEEw}EW4>gCLh`2O>PK)gGhe^zWm
zyf$n=X8%aQketzsjh6wfe`ioEbrB1{CW)F?C=86?U2m$cd~bRxOXstI^ONqOM)wGC
z*}j~q>Bn<Ze7E3Ay&VvHzaHCX*b?_uH^DY?C3?4a;z}(QY-_lQws%DN9NbK%^mRT^
z<5Sf8q&~KMI)sTPfvo6bu^?u^6Alf3N6Uw<(5d$lmiNgsZ|eZ4YdFBaYqnuPr4n<~
znTb`m5Ajatn;0efQQ)Z13lYXiT+US^oM@Uu<)bXnbKMWvDZiSO=3Rp~<(Kg9Z5NU;
z<{+0RI)=%~p2o>ZN6|V&5;A5gF~d20Cw;pzu~I4q-CY5s=~pra@eaCId@leVoCTwQ
zQ^@fJ)u^glOV^8RhATojCYUC`Kdo<Z-I#P3T&P8D^rMN^xXCPiQ3#CuS^)i0+j*|q
z1oB3)4K^@2_7abx1D{gY*Ixj8)U`lwe-sXSo`vc+4WM=<92E*R(eBa^?z|_$ENKCk
zP?bTm2iBlor8Lv|nL`FO`I)Qj1S`##sgN?_Eo2IlaC7Bm-0@;8{Ol@%^z;NwDENho
zJ8$wC<?|>S@65dBsj&Y$wRKLo6H|>-=>OILK6eqp{lQio-XzN{4BW*vnY^L1tqfV{
zGkNBe*#v3-ogllb%lRzY2h<cyWFFD_Fz4Y!_E15FXVRY}>7tq7Vm^zV$z;T<qZsVu
z!qB_%FL#G$m8Csfgx|$Su_&J3oZaIGb*=g2#y)-)RC<km^3FuLZM^T1|NTBpl4m3T
z+<>F=8_;^BEb4DvP4sU_FrALk<YreH=xg#EoEs-Ny(|^xH~vj^c+@Fa@nj4WsnowF
zuG@*#Qg5)L*MT_>OEQ1ed~BYk&N2@CM1ViwmGvLY(vf7h_wR$jijhQdx)uxF!_QCS
zUen}%PTZJtt4OTrO!8D|3+q1-MgD2bV3M*fkQ}~{ZW|oM9Io7_vs$NM{nY|;xa}z}
z>MlW-GqTJy(G0S-+Cl1qNu+v5C8ql5v*i#&B#o?a?ygK!k$r*+%f4_46;+&^!6sC8
zKhL|N_6ZuIys&)w4t8(ud-ytGE8jg|Nf^IZ(RZ7PUWbZ6Yi%{>TTsGx%jD_zf?-U1
zB*q-wvgzRSqu8RN!Cu^z!r;goxO3YdsQhROqip%E(M(n9S(XI@fyyj@Y(8ezr3r=g
z|Dc7t2DXjkm~Plx`sPC}&I(L}DQp(Jy<&p_%RXaq{4MN1>_L*Z%*BXlui$C#At<(O
zgQ0Vof>&P0iAkazCanB|Dq|KAk#@PO>8f*Ch)5h7o%xBU9LAB7alu@@HSe+yT?S~x
z&&*c^q1Vq~zO#D?#l#ed-!RXyl(PVjVhZ)Xr?70}CWtq@3QsRRr>57x2|s4>xgu%r
ztCo8MIlIo;M6MtbCF}gjrHl3ST!l0#*j)+9+84Mo_iqA!T|T=JCdag0CkS2zmZHhA
zCwO)rh2Qak<gt7Orax35QK76l%FLaLjvwGu62HU7uYWMOJc|bH>Ozx&R3fUVK-OxP
z3FY_JqWK(tu5_&ik9m|s`t}#FR>qip$ePZA(>Z}|vpVZ**^RkVqRIHDu{h3sD<-7W
zpkdNQym4|IySJHfQH>q2@oov3UTDg!Wp{FxGgq<VZ=EP|Y(up!?>Y`CsH0UpM<?X1
z4oS*v!-k&snBEjZ6dYEu{dX+Twc`d#B#4r-JIPi_Jh$<$e-WK&a)ML8(?q)+Rk^h*
z<Irs8IC6CTH4<F<8pnnXg4(8uG&jnYbM>AEZ{}^l+^2_7=}iX)yiG#&K_AjqEK0sR
zi!$vAv+!@_C&BXnjPQD!Da1J^l8Cqj(ru|DJoz#cwc^IJ;?Xtu=*Jb@@KqH|Y9*k&
zavt$`nuf2dq*(O$@2Itl=d@p#2j6z`KF2Im6z$%?yLz%A=T18s@0bGqIk&l&?$v@j
zcQ<3?^Rw7gEl2Kblq0mN19c<5a8?P;B%8H?S)>>G?JMP;T#N>%uYN40*#P4CXW3yr
zQMOrE9Lrwa5QNFhptmL}a1E-x*zM}f<Zn(P*H=;8@gtsG+v<p)?aFC;T{+I+nF=oV
z-(elhBqfKRaDSggB8!ZMC5qkLgUn-a#oz*H&urzm-rM{+VH>&ZdJaFQrV6t6<f8Pb
zK=OB=7EN8(O3O9Ik%%X<Y*oDkb51Kk`?Ot9Ecy(3?}ec9(iVD7&x<r$>5^@CjgiYy
zBlJZAjWkW>&K&7Ru5&dS4~=KaKCij1$V(XY<2<!dc!jbVC!l*;nw6>3B3euZ#L>PF
zEe&g`qaIbEb!LDt^63OzTlpN5^PX_-mohQ>(iN)xnr9VmIF1#M5<v0A5nMh?nZ-Pg
zBxBYZvn!Hfpr%@l?k1|RY3f(dYCnhP=Z|G}YA$F!V}SeQ_6r0ZV#Ls<1Lp+r+#IJM
zJY|0t2b7aw$v1H_^*{`!rj+5*dQmj?=tRFuR;(~Kncmnlh&mgKXy4gM+=0*o=zpgb
zQUi@}$moS2??yS-Q|81(D_kJ7UZ1`B=L&LT4^qK<6>{H2hlni!@>?d9G&mlH3ooKc
zp;Qxh?)*daUN(}|?VQI>-#E#A9NUSJ-L0rHe+(SldmMgO{>Lep+(VX<1}>(~urO{m
z>nauEeNz*Zp0kx&?!5yy{1fS}vWdiLi!!JzZxn`1s-W#p44D5^QC2^73uGnbljl_c
zKU%ezMZ<EQoo$B)!f&Ac`%j=dlR}?y9BIjxB9AVJvb>KBT|V%^>`{;Lw4)1a@mC=$
z59^R&{{WnJ(TRm^*^e#evMis*;)U(C==eyNY0e+Vb<qlZ>nl&D8nwdC2XpXdRtPz&
zv=qB%5l9U_i>eCxOe4AvS~lw8yccpzO|?%rY)xUYiUt<f9>6H~JHm>k#rP*$o0-as
zQhUqE<c-W7n&0pN4$b1&)fg?hX3`j#*2OQ}C%wgHn-F5`!Mi`5ZE<q<Qa03n53^T}
zBo6CWa5d3VBrZmsd@(tMajSTaQgb0v)3e<6RrQ#+b1l&uGaXuLvyppN$jKLkfzGvV
zOzs%Ne&2bDQ7N;Te%TIz=QmY$+%ODOj+wG8YxSAf?p0`8Sq{m6OK7>*bT;GkC$8#Z
zCVZ*lGpFW@Np}syUoRZ6ME?VHtd}89d5-XF|2X#kKrV)#ZRB~~y;kKlYOG)1o`l<b
zAjk6o^E?fOp$GR9(_7Ze>f=<DOZmgq9v%`5@tl?`T^T5Hxmb{0d>QGv2Y52?1YCU<
zi}Q9Sqv);^_%$sDHBa&L46{j0+j+Y%p!f;ceygDG{-3QBci?l}2IQ!P4l}O~5Y|Yo
z!HY{QSih$hvFCa3KY|4OXE>VqT90CRlllax6@_Li4svehFQM#hF$!yJaX#;Nxota@
zXX93Jnbw=2&bXEnyQI(7^LwRp8YX0)Wj?ei{D!QP_RRda6?t6k&+In^pu+hZR5^7J
zWo!J2^~Xdy5F|;;JzZGWdQtp(-UzOeJ!p9D5qI~)B`h2;gWG3j5x1FVahFmC#Js;o
zC1<MPdVw=am|dryTa}pH$g%8q_I12IDIOQ==#%yVTXto>D$~_n1Ho0hI9a`N+N@>C
z^;_}x=Pyx2EdMYkvbhj8S(R~`eaYxDy%o+h6_TgF3^4DL91H1K#@NbJ7_%^yZVulH
zZT!x%WV;WSbn7V<dC)NXx58aiQ++9@b{Waa&z3;<w=>Y}>QBz8S&|tv4<}Dm<tlhK
zg+C+`&TB4f>NY{0=@U^>emh+5>c?dJeE9j-;{S6hW_@`dPF$A?OP8w?$FaWLESvT4
z=^KNV*Zwr>Ln79??-V?3kYh1P<uF<G2p8($4Dm#gTs=LS%^#@>%gA}mY|!A29g{&Z
z;~}&=HVN1F?M8=3GOWo$8uS7=yz4)jFvVjSey4;idUOx2?YoC3ek~;u@!#m$M01wR
zSrFSJk4WY5vqXP#AC0rvPOKBtso9r&suDSs-8UIQw%zt5#wYna>$)X0J!c9T;XjQD
zCx~Nm?j#b?xreA7>=A}dJj)$R?4|Z|RoFA-Ni45CN1*?ef|V!V$FE(5y0Su)T3*Mw
z4~lR*>L!rF(DzVVcb%K9)(3xD<w*04vmh2~1Xespp&(}qzP-7e_ie=D&CI`CLw*9$
z-tq$O4NH<*>HtyAnu4>f)u5Sn2Bk+F=H5&W!t5F?EcFt>B03g){?mZX3#-t~s0WSR
z#JE4wviLr0GU>R;ckFK~vuaCqz$d|Eh@X>|YcIi9uT7}$VtcHAbxcq_pv@wM*6feh
zXKwkI^RSN0#1Q{PH2$g0)h7GU*Q=Ds9=GSb$4!$3Uml=y4!RNjSEtDNUmtNt<ZX;T
zUxCK{nLHbew^gA9sjXZ>Y_v-;wc>vD{mU`v_^%(;@28OGvKP6BJwklCz?8LqNXG84
z&Gg-+6>P~dT^vwa00O={Bp+4|9qmhKj@t~j)S?5P_gACmS34s2tB`v~>(HxKnRR)J
zlZ03~Hs9SBKR@7el<Px@$ftuKbVxz|RffM$^-_;(GU&Qj$i2Igi7EFqpf)a)t8iF`
zq$!(lby4{1WCxZ<*plF-B4{-71bR7HF=O>@-0+iZqM7mo+t1g4qv#Cgsx_aU4?aqc
zO>CiQWkPh|RD=)jjltHs5p13j&p|bJB4r6n1uJGPBzuzb;nKZGJTr0xn$9^)YTNi+
z#v3CNBRLtXS9w9y-v4I9I62n6P8lO%AucnpU_lF4!WPdJ<V?8}vkDwdTucXHp4<_-
zg72hS#kKHVZ6!9b#1pqTDUuZ9!<d(?O%<y%N$i(#j7Utu6P7=y)!6AQHDxVFyKhp}
zquWq3AdDPJlxCHRYD8W>famfDV5dSc%=jn|N&6{|b<PBP{%%)lBhRM4=mEFc>bT#h
z04~I-utUCQP@-l&Iy6SZ&|*E#aq}*`cYZf){${~h`aOp66^iI*aR5%eT0y?AdJMX=
z*OGsMwGdTn#BFr?L;a6VVd}fxNTS^z?)URFuIJGvu5qt6?-fHbrTPk*Xpd(<za8Pq
zW#$vp$J-$5-v^wzT$VK2T!bVd&a@wm7dXc&;69gom?0NWM6al^CtLZMRHi5Ksk4B9
z&w!ThOKDS+E?QTpvI3qlmf03cN+<8&Cfq6G+AbHP(cnbhr`HQXDpyc$xfF@|tRonC
zTb7Nie}~7W+Q8Z-XVM=&lN?M@Bz0^AdxfHGWx;#=!@sYl#D9VvJCq<=Y$=hS5+v;7
zvQS~EE+;&+oAkxr#WX#2rnraqpKRxu5d6L=VW<-qWX~eMFYxc(1r}`2nh;XlH;6xT
zZ6HrYOjx#XI_oqZMTT}d!@SuZY!Pn5+UnVyinR&Y4vd8|HiwI>7{ai86Jl~}8fmQy
z!S2_4Fk#&p-1>b2D@zL}Yu#rP8CN?dnPkVNg-9?P{(I4H6NQNeT~PZp&}#Qj543k3
z%XS#nar4^t!Qc6BdFD9JJa0;XC_5dJT>b#7>{gH=o^QPR$uPKYX@`)-1<;z_gECn<
zY}IHjxNzALwViCS>v|{#1^l41LZ#rznhaw7Y8-7Um4*FN_^e1;9JbgUr{_IB(1s89
z(bw1t8#PC;9+{ue)~&@e4dj@4pbQiHqD(5s&myN||Dg5K0J44m84N$UgzQ**7Upbt
zjM3&&82C+<@6|Wsp6{dCBiDJXkMg^u*O}mVW;Y~FAHuzF?BGuAZj3opjmbx6V%xxN
zm~*WejpELu^i}QZTG@@l*QEh?A-4(>H6qEgi{fP6dJ$GOSA!XdH6Wd@3rf7Jx=^76
zzuOh#SZ)s<Tp5jC8{cuQ=s_#;ijeMqfC`gl(md0tB*(;;4CW^jtLUxdQ}Jm!V&xK4
zXgrSc^Sz+;?J@ZIay?0zItRm!ilDVeDpxsS1grSA9<1wJY1b81c2d)jKc6RJ;;E;&
z%kd}n1%%?&)C&-y><NEwYz3=rPGq*CJk#U**hXgesEl$q-q@?pPVx7Qp_L1XlEq{E
zZ94<JHq;3Yc`wJf6PlEy?kDGuOOU;#hTN5##^jT+FDP}D;oPD9#BW9o*M9sa?i$y_
z)w@-3+OO8Y!~w#@7w;f*Dn&50Zkf<QMUT{G@tM{4hFtc>Ww`EE7IbMl;u5z}q)YN0
zvMDPuYT|$RoFeW?8m2?<a)pZ_m^f@+L{(PGFnftS6j>fmo4%c(+0!=A!1!!1X+q5Y
zHx~*gFUB^l^Fo#S1*Av60h+r+na<0tI4Y<DTqnk%^g|<(9-xL#ofX({Ezh<#dkG1-
z>!`7gI9ok4mQ$H?4l6<n_+53daMN20vQ70mm$FHWjasuAZcSQ>S7S7h;Z^h$$tMb~
zJE+Sqf?tNknd-e7eB8_D*B>jg@HbBEN~u4FnzUlc#<3)~N(48)x{VJVJ8`Z|4*V_B
zfGdL=*i~a|kQ%3j&lA3JeG&b%hUdLKEa0De!AH=cSBLLE6vMb(hAhGLJT-02!`cfY
zxMP#G$jjf3xZR5|nGzpjDQG}*mu2kTZ2p|PD4Sjl{U9i5G$%{$mEzpbSJ24X2JUy;
zk?%!_vX?|jX6JLxfbTSw{LJCcMjp_0Sp~wk^0~ls9_;qQzkHTkokhH@0fnXK>BYiW
ze6FcQciFGUTiqW7vtMi^;ul{-f2;vhPs&8a!y5@RX~fhS+0?RB0~drkGKGqf{C@Zg
zRSaDY*$+4@>~6w_B^!y<U@Gal+0WIx-@<<dxloWd6JNF(Vo%f}w((>jY+K(7Iv=C(
z+_M3$S>J>;U#!PAJ%464$p*{D@r+@yBiOxV7U_Y{oc~;VyjQ_H3B^XiPniOIv3dtI
zwmd**>4W&*(2O1C<Vn!qQrK7=fy<iLlE7ZxhiV^;tLM(aY^73ImXZP&K99npy_-?p
zSq$FvXJf=K21OraSkXH@5H+2NZpMRfMWGdfcCQDza0nfCgz&w<{;F+Lys7oXxj4;Y
zJd=52h3{WUu)neaT%pxvI4`l8WT$V23BT^*`<%1n>^e#OBD(++tj_Xt=J(X~_G_v+
z%t7|V3QYRbf~N1oIg^=d!PxL3ZXpXu*8U8xK=uymo!E_%ujjKYl7O8dCkf*nD%Lu0
z1pA*x5xEx)Fm>oIlpR$h1&=k!78@O`ebgZcpnVuQuoP|c%xHmy6I-=<Gpzm#9C^0@
z_l14Mq%rmwz0-tqxzY&%ud>j4(O9hSxddNdGrH058Z|ui8B;nXxZji3fc)nenAKy0
zb3=C%FXI~gRrrJ#f$zs#%MO#Fg+ZL&p{F!+(-3SET}pyyXb3{=qBs%0LmfBD1W!)2
z!_<9d^h3TClZ<p_qu!jvt^eL&p3P_;*>eXccm<MORF$Ndzo*khFX0Pa6IS2uPoFCD
zjI+@?{J!l!+M@2tN%8#7!JsKD?12*NSuBUv1#1LuIqoPEXGvm49l`I$Ysi9#iEMWJ
z1a@X5?|nO&fMP}8sD2*L$a^qAUk4n5lI|6xY}6!S=wTi7v;0FP_r0Z6#X<0)!G($O
zx%<U8Z$j6a)BNn_9y*UNhr+U9`lD(B>oZ8iy+ToD8$XizH7+CCFUF8pO<yr=Sr(z4
z@<g-UhK&xK#)3wl#N5ALcuK>8`8GVIzOfd>V@fqdNjgDZx*3|ip3Y*`zQOmPYABK2
zNz%qwQIU)SxYvGwvpOjc_CwcD^@l7|{OOK{26wUcOEejklmt2JA7I*Co@3cFf?R!(
zO$_cWV1p<I{z@M>ugmSQLRN<Pw=3cq(*+n8{2fXf5+Oe;4jh7Kvdqfm#ICmlYce;I
zGEINDH-7@k1hwKRqv>G%JzOX`*^^8mlbGCEStk26lPYNx<F_jZu<t70Z@T;d`2hh-
z)vp8F*C**KizYPnNQ7bYODNQhCe3>}HaDgQwzu5H1nn=vL+cOF+Q$=+>`mlZCV`~f
zPL-V0kV1dGOPHi30y(27uKxHHy;^@`Q_4i9xx0XyR`(J|I9s#kHRUKXLzbj=9wZk|
z4WUc1GZ{Nwp53S%#WJf$vh)SNk@aN4*PJBIb9xGnHt9vzTnV%Yi6x(UU6_KwGz>1b
zX3T3WUHn*!$(<Na{8Ojmw2`&g=$8n!<3w2h_rqk1=xuN;Xouj;k8nWA6E?kCj@Gpc
zsMVP+m?|&9ZjG2j(yq-Wa~GX}kV&cNqAB3Cnk7i*pguVl=g+43nDV?Cf~u|U^i--V
zto<j?k%yz0W@!X@?W)d9KNq9P%g0zTB8&Wr+YUSYo?^%*J7#uw8auk-G8Nh{Bdrhg
z*nVYCQhlQg!o2_DiXX8!&F3sWk(kP?e-{Y$OdZV<HXT4k8@_YKvq$^m7vN0kX-xfF
zCaQ_gVXodcgwOUz<JFsL(6FW#Rl-)m<-QzveP}zjZ8*e<Opk+Qt|#c`Wvh9fcLYk>
zJCN-D*(|etIn}9FWA`5NK9cD>X>ye}joo+(xZfM7+=GeOcQF$)CO2~_dlu0<`*_cG
z-4gbRXH&^XG-AMn8XUhwj1`}^W|Oq0vISu;;nbDuxQ>*e<<TzcyV9NwpO`?RLo}G#
z;y1AN`6gUHJsK-rE@QLu9sJMp3@s}PCA*I7gj4F3sB|z9dzv+<OUy=m?_tHwn_P?|
zu6W~whLNc7;UEc$0|w>hY`qPCFKN*w>BsmE%&IROw^E3%eisEDF?PiLl@u-*6-^sI
zo+Lw&_XUNu>2%!mt=L>-2REyJaYF{{xy&QYDA+lMiabAne;+s!pTPO#G@tz%dK`yE
zWiy!1ac4NxeiS0LGeD%{94a?9!he@^$yZkcA~7`%6($}Qe6P^Q9h?9+dOGnl7;&aQ
zCYN?^wZP|{Ni_QQCmQxb9b&hSM4zpVu)wJlM#luA-^U&FX;&<kO-!d;XmoYh-C(Xz
zbt>sK@grvf6xrW*7iihSU3j;l4bxWZu(hvJ@cTz?vP0J#trjQ=R*GwqoRSQvf3y>v
zj*ex|ZnVMuw#zsezZkV=?4jX$z9=1oaK51wPAX}WCYcRbG5!HvQ?!^x8SklfU#o(1
zCg`#zl`6Wgc_hh~ZK9tii<4U&M>rFMbUYL156*hEkP-I;tQT3c#{X>iInO>!4_`vv
z=^)Oie}Lop=go&B(kwCWBnf`Wu>!|+U@V(SXV%YPwnZu6z@Lxz4S&OLo9?3b(~q>}
zpgA5Jor1Of68K2j8@g)dFo&x5n5WzTFaM0fWEjcv@4QBrh1UTm$l|xchv4+UfpCTz
z<Z6Tl@xI@JV;)+w?R!;O)Qw8cTIzyey~!F>_i`eUr*w$i_9H~h_9Hgm5n=0+9l)OC
za_ei%pjSmeEYGdsEbio??<Xgsw0<Tir4R7~TS@Q=Zxt#Zn}Xs>r=h{v2kTrEaE65l
zS-o5hqg_j3Wq}NHi&F%xOR;p+t9D#cV9uhNZlMaDK+*<}a?j=8@O<OZOglAQDEH<!
ze*T*!SYOag)q~Ypt@Lhmexk=*-rt}K%VvW^0mmKK=FM~U5agzs5Zh~dT({kA{Jq?Y
zy%^d<oHvGZDXv2BUt>YE=F~v`-ltrlu`-*69;`oiEp?x-giH6xGo76++%CToVq54!
zOy=eBbCiW#uJACrX6uuW&qIj(0clu&X&>Itd`~a9hjXt)F5tp_kD+OvJl|{6WF~%M
z*w}dr`vbQU^TsZ&|A`LiYMsxlS1Yk4jNw}4sU$(A538OZ;dz>dY<0jeuDwq1n6fd6
z=KUTeZkxExZkq7s_eTsdInHx)ZlP`LG?sf-1XBC%3A+|nQv2dU^v}P>_w9p7lvqwR
zc{L4_zdXmUSL?vPZ#4-X$9LTqL}RT{IKK25g{kjeV3Vgb`)}EIbRX^%ekeE(w`Qo)
zk^==?T|yU>ZTlloi7+L~Z#f!U+JRkf9#HEw`GQ;hP3U5L2TZq4U^V}&Nw2;Tt(1Wj
zrQH_}9aQ9+dJI_k=wqll=kWg-DuNmM<uv(aG|Yi4I5Z`O9{f{+#)ZMSS<?vCu0KMh
zii${zM>SgcmC(56m5|5sK*{nhZZn$1T7~=YOIQ-*XFj8?+KkSZZO6qgdr)6LQ4m&=
zOowGA;V7?zaJx?$HrVaO!n$>+tZ&3#H#bB4`eCRLG{B0PrbKOZ23MB!k-oV-0~M=Q
zL-}5H%rtp~TLV?eIdL^I-PZvXzw8A|LvN0nO5+}T2Q-%2Nz;EMLFGen-s_bQIlnrg
z+hCj^A#^u(uC;{C#d<WzO_|JIV#r>NS%(kWh19idzren{1{<P`P~oB{Ds<~G9BP4O
z!h1MgEed@fTEI?)6)3-QHj#e%P|#wa!{3>Y;}z{y{AV$MUhCddm8qH77plgBpcLvx
zRq;EiCm3#OLPQ-z$YI`v(6aF%e%T?*mW61L<5gOyyi$~<mM%jH>u$Q=`wG4j(coM)
z4_Mio#PZI=qd4KM3g+;8#|y@DVFI7yFx}xxMopJuOCE2>)D!&Y_*tC1O1gylZnd~<
z`DkKxtrHx-)neIwDNa1fm92jf3FZ2iA<eJ=!>_->Px3EO>Jjg^nskIFck}$+#$+7c
zEXUotor-pPHW=A#%X)1r_*~u;mO4_2ncoX!-*x4g<G^3o{>zYb1z6$-{`?WWL=$(8
zS79Pj3D+Kroq)pFzqIZY?=#JLf-`P?<@}$Y!@&pdpjpcl)t2nTRiipUf9*aHNp#}!
zW=C@!XJ<0`)EWGl%AcNBJc>n6p2GFI3cR;%2IwY>k)i_v404oW{}%AwwrW0`Ztek+
ztI}}CQgQO)#3thG=L;)kW)ZLKhr+rdV|J^dmIfC{GVdWx6#Fy@=SZ$Y?t2NBZezeK
zhpupL$T7a~LB6y&W8Q8_t~TX6{Ea|-uQY+JDE$Os+y5XPQNrCCca&RNZwG#t=aM<E
zdN9xRgK$)|8PlmKCFwGAKu)cabB;D3Weyqe#Dgc-Jvs#O2G`NxUM8-cSB~K|CsCAV
z``G4`P_>MmT%5^wJji#;o-3$xmrUN_XBr7E#fQmeX%3dXaYnyseE(>)0@=9xAehh1
z;QnrH!v1|RJYOwcxZ++4Zj<7l=d;q$aljH(ntp=sr(GlmyV3lXIGNp*O^#F;vH+P!
z;Nw09X774TtyjmAJC|%gWyB;_zoHGA-&fF=eG7>BiIIHnZ>YMbg;Cqw48D7R7)?B8
z!5P2#WdDsqblhgl#tp8)qH%TTYPANm&(%P<>2Z9gX+wg~zlH>zS}=*|;<5&Ilb&TF
zaP%L~@Y<ORIiJoD?J-9LwPS8_9nX6~{>}!XcefX}tGvW>lSi@L({_+{{b@KX?FrjK
zfr$;8Fvr*LxM78na7}9;saTr`8IRAC&f6c+B489Q-a45%M9rh-d!{k#n;wK5*h`!~
zHsRo&XmED&qG2JGRC}P(D({jW?H+4~{_}PSdAAC?dvF{xJ+PlNU-iaDp%sbs<asgL
zVo>&HE^C-RnmA3<L<d14t`och<x*u%JfMhHtVsm2`xk_2^ZuQxVsJS9oX;i_a>8l=
zxgKeH@Uk51=e-cVbGBf)Vj}N4$>GFiGr02O3}!hSVAkgwC_m;fawA%?3j#=#s{#g{
zs=->Hr9$Oz#e$RqF}^E0j^`9AGO=-ftniN&h)lnYBJt;DhxvMAzquaK-XTY(emsZ1
z?@z$Gy<fq2ju~c5nMF!%_mIyqe0~>n;c0v$1`hJwWy3+-#mdn;P6y{+SP1$`7vR@L
zT~bl~OL#$5jpUhaf#lEYxY+bbEZ^-qJaQBe66DS<oSROvlSIhp=V!1g_aAOAksv;N
z#$v+IFDRL*Mk|&Wv40!X$TrJ5YPK*BUp|`5&O{xD7`@f#uyO<^b4Z`K+3o@^tAQ5g
z{Y24+>U^dq4t<R7puV&|hTDaaj`p>z=W;x$c-n;DA`_sn<{|x)s6?`iE`iSuPa4r!
zhMnKHLyn&<5j8O7{D$vv-IHeuPuxkv!Iq~`?{Je}_m9RAvHW+?N1HzTx`<3xiXqCY
zBQf-E8+fH#!zZ3~e2-^qL>k$!;lOh||IHB3yvu;%%c8jb(S4Yi-wfToD~U+y-)kyE
z!}#xkI)1)z4yQk|BpH#{Xx5BQ6mQ`9x4h3RcMAodg=V1tN|i{;<-(l*yioR!IWX%X
zl)TnKn@&t*CX?lfi2iA?`xuR`haW-Zr5;!t=tUMuU&OuJw1`lt8e2A{abCG8<evC2
zXT$fqw+SssmK)FaPbkCPQ$xr;$&s7}e>Z<PHw+%~UOlaNc^q{p2^xD1AxdF;wT;O(
z%sAmmdU(EC^lcFq#=jR@{wa|=WBx$>t?MMUo6sH@$HFxEJ?YX&)+=GdynPkfaOYEu
zw;stvN=1cr(+fHMgJZd%&?NF`HlJtIwP$Z8UxHUZm2h}*KP=Ny!FBC?SF<n})^`Mf
z$(uxc|FH<BK^i<v-cGV)+F-naD;vF%&lTB>WbY=NK#{UVRu_3L<7e+8EZi3(jGBAL
zs-2$~^|bk7VBSjfO1%b}y`G%^^+1+-b_>y8Bh8*U*uhMvR2(?&k1=}-iC@hc!KY>)
za%#pFu*{GXE_@XO{q8!fk7qzWKQRnzs*NyrAQI)IzwytVG@@^(O=RGlz<R__&ZRdD
zv|UECr7aEU)jJAnjh%(r#}ep5r`2eC;}_@gehc_}q|g-$-$VPn$<)2jj$DkNLby90
zxaAhHJona<i_UDudw+uP%Y7Mktx$oj39P4!hqOtHdoT&|IYetk4Pe@1J=XE$9hk0&
zhP`8NK%G_{&Tlw|?q<ipzivKL+!#Ou6~2Ls@==Uv@nZdj8j$~ZJ+?>=2vw!u;;`>M
z=)69k>0uH|zr0N9?c#~U+*TB0W!QV-0Lr|QXX8gl;)3~BETlG$lhHpz8k>*tUd#f_
znl3{g4k@us-eS!1h6s11R*(HlR7M#P;urB$62_@BlVU4~3HgdG$<sM4KL35(MV`Gn
zX+R#H`h&3YEN(Db3+BzM1>2qnknx*6*#)0QZey1$D^gG5XHz?|yY>*6d|pU$a&=jX
zM?BHLV8}`v14;bqsd(4Mj-+L{2=1M_L>D&l{L9@!&P+s_MZbKGj}qV0LD?htbNL<Y
z91~AAU;Ym(u9~tk|7TzncOCP*E?9{w8bHeALiFnr3eK-mWNGyt=yyPb`B}UWZnnOL
z-2s3vH~hiJBd*}9sop53v6>4V{RNGKd*FVp42#>Y$}0QP@XhTw{A~Fd@`uXcz*Bw~
zP;wB5`Cg%Z#zxFY8^x;O4)2#c%q^AUxt+<MsKQkPZo{JGJRhhUuI<?X*IP7Uyz5`o
zdie^s8E+#s?UnG}z6*EU<uKhriTRDlMfbN_82jlkES=&+zP!nR{OvRNKi^5zc<KN#
z+uO0w?FOyVI|=GHHxfm&M|eC#29uu}GlvpMQs^PXTjO_gW)q)c-|sV6H-+aHx%_|&
zDT~ms&=vRe2Gi`Nl01v{nsEHIH=NnY(d-G&^qgwC6MtCm$34YL<mr$*k-n-xH*-f(
zQZE&!Yw_&!Tz<~<NSRgX=t6X^9%R1x0}{qEOz*=aAsIZ6#;vQt`lOFAOtu!(9(N1(
z)OxbG&DJ3DOGI$F;|AxkwFiYO)rt1N8)1*~R9Nky$kf{=vhtfhaNH$ta>sBi`8?pk
zNq5d7?Kj(n7tBX6*ItUXPi>&W*#m=nvZy`3XP&uj0K<Z#(f`48W}VPM^CqOz{)@d3
zb5xD}N~92;{|ypy8o6z4c~IVX8l%@1Ve1J#bvjiFV&B@3%+?ktia#m*A=8Hvc$@0W
z+vCfl%kWA8?`(J)g&WuEGWTD*=t!H>=zFk~Q{jJ3|M~K8^9P^zUbC7w{H}u3$|YDm
z!5o?@ZgS>_9P#*%a*Uca(<&r>pCD|3GAiVqvYK&CgjEWk!kfW$;PI}8i(6X?HtA0A
zdU6)Hge&qqFfEcLZ-8!(4`EZyCT!*J;48atWA{1%L`|v%(>)rngN`FRM}5Tq;`Q*>
zG)2Kh&*NYy`hhl!PlS-*kt}3qA>7aX0jq0^p>yj#7@l0fJr$YEq9&VS^tU|Ro#RU0
z`j5x*MJ^b!c^!IYx8Up7vsiS{5188Kh}nx4vUBGSL7ko=&pP0DvB4GSJ2i@24ef(%
zw>FWGF-iD%;UKz|@{E;1F`?Q1muMUKgPYU+hd-;1B4%=xXniP@tn`;5jlD6D7oosZ
z>k^3HyGMA~+6UK|oPmq4gCXl$8kFb9<39DT7*Rh3UN5=^;^WW3=VUwl-8Pc%`>7Fy
z!JQ=N<6Zc%VLVH<ufV?AD00&x3m3y;SW_I26PhI00q1GVs^>5YCcMPZ>IuZp@gg2~
zxQi7IXR+LSJVu9Gu-2nMAMjj~W5&)*(LkFi1**~L-)>B-{wVSEc*VO1CD;}-O*TtY
zlS!I-kx?)6=`pt~?zjIqrq(lsmE1c-3uW%3&#_ZzTX+{BHiOQ|(goW`<A@l4PY{pN
zMv=TAs{lSL;~^hL!s{z=!C)eqGYz_1m`wr$tgufdfz-+$M1?IeLaQG+P%n9i^PQ~1
z<g=1+#=jvr6f+g`!qu5u!D+0nUW|v=ZAWVZWmen&O+eOuz(4i!q;1_GrY0{Iid|2K
zdUgJuyZaHWxhYMYj7H!sBR)f^_mFG7^9tIcG}u>#1yp?%-xcrLg*m;uxzfwBu&w?O
znZ~J*3DL*VCVL)B7$L{brHhmBWJ|K<KsN3i(}*W`mBGrtB5dunIbeSD5@#*G1bvj0
zNM*hWh==g3rko-e`tcvco>@qO*<`AzB4B>?GCX&EtKidLXX4xwNSq7y;O935xIXhX
zwni#&U6;-iY3>g+=E;M|q(r*EFbb<34OvfD9QF*q#nZdT5Q9ts89d3~$Divn$&#_;
zh1dt!b$bp`_H`i@wg<UKwsLHEo*J3+do|N;)Tn+JJDTud8;lF*-FPL#puPVJJWHKG
zmPU<)_Qz7}>*Y;wtD^#})q<$AY#LrLTM7<~3dsB9I5bg_W+$b6Nm$_$a(=;UuuRPp
z4D$P)>k5O=?EV+hj(&mc{o>@_onFqD&yZ#ts<69Dmk~LAYb+l~;#%a@nUSp#3whK=
zPelk=M6(=|+q#8&``>YJyUqWeMvu_M@e^DRIYb)mqA<T_7btC7f(a@@;qx)tOvLAa
zm7ID8e6>FV=Iy<lr2P!SpXJaK?!lYGOZmN&JNy1e6dlqdpmYCe7~uOSWm_L&qKz5q
z)?UMh_bEyo{EE(ICd_D=63#Cw#)k**gR>#;o_rFAL8n_W=W+)+eg`TqI}5LF&xO2I
zo3NvEAJ%Jsp(TrAINjJh(AtnK@Yzs}T8Ry4H$0P=&szkF%l@Kix+G~bFk-tu?c;RD
z-G=1xSGbjl3K%`!73Y?g@i~ko@VV5LyP&Q>oJ>1mPFxk<P@RAp#D`3)mcWXl%UC{@
z!rgy)=&?2f7d%Ts<LkfRM-s(4!z<`Ly$V}<PSI?O7CJ|MIg`=bOuqV5V)F(Wv~kG=
zpP^V>``npJ(wI*YTpfx36I1-UC6pK;@4?&0JLv_#=_H#Js6BX$D~!B~Eju!)`I}7m
z@1HiC-}e}-`Eg>=QhQ8#_?{-$D52e~<1or}4*BHaj5eV@I7jywT2|&)=WP3n4I)RN
zr9MWOYw1nemu@1#zwXeTV!Mzo`~+rV`*Eg)D%=Tcp%Q-+A?tQGDg^O8_fS8sBw3Ux
zUy9*OsxP3*r!N?tZpHf|R7h676L?7XQiy#7T3tKP+iwV0%?l)%iyhhG&9-RJXGd1m
z-@>}APOisn85BIY%GKBP;!@ScL{EJIv1tpT8b-=&_k<51IpZRf$)yN`UF5lhN@ePJ
zWiwuFPzN8KQdBaVz*Ke};mX$Uhto60Fv0iZ=+GKOq)nd!%W!7e-Uo4(`bwzwHph~`
zlkw#qb0(TC!7?POY1_Jm%vR9_L`nmz-_LO($tzcqT_+@1y$F!@|BCV4*hlorm1E#x
z=gK<2#?c<hJXF3=%Snq?!PihZa$$xSF;{WtEW=Lneo0I6+0&M)IW566=QokxdFHgU
zO95|0tCJy@w=_#M1H^oPLV!pkWFB3>bxW&aMtD882HG%@cUjj=EUY<o)j!xcPMuwA
z4#fZVj$;=uF2@@ePf)LLAMW4^7Zh9e47d5eKtI1_v|+P2IlomBD_cWg-tqNN8_<o~
zQbRP~Yz^w$uVnr9|ESg92Sh432Swx`3!c}lgrAGDiKX%)>Yf@eRJ=KfT{`_5gIXQH
zI`={KD4SvKp4&lwcAbHXyz{|Ssua#{qIhZfBzAV!M(}^2$l5Dq*~T-X%wgVD`nxZJ
zm~XWrHzyl1Eh_>g&Aj(x%5;))pJPWGu0m0y2GkFhqE9V=N5LLaIOYz|iCX}gmTQRK
zq)e`S=}6{PmMk14dKz!;G)5I`{vGCj4AfJ0!hgl;Y}Vr}6g^Oi8yqblahU&K>+}W{
zzU%va0^jGTJckb-wnE}59nR|+Vejld!stEqcy(R|4qc^~=5~dD2S^dI6&qo1;X2k<
z&wB|}A9H%I&vLC#^>HX|3m2+#lV=fUW6mrIp1U)Rbr0>ufVL1EI8zR*#>z3*<%KxZ
zp9B`Sc{eqm>ozg$1Ps)Mg;z$ht4*iK-OVGQ?0~4S{Wqi6Y>v~H7cAID{<;4%CkG8b
zWrJGPQnF}-2)kICgg53-hmzuroXC`M)n3aF3Z&z8@k#SEa1ogbsrQFb{PPW}w_S&Y
z=`RE|{=59@uN&tczZXLcZ$QVb0kjp3CmT;B2pT5NBms4A@t@K?TJdx_#$4Y=q<=rB
z7hM@dKi^BelkD)!)@!i;^;guCS7u+AUBr}iQ}J2MeylY-&S@{M7Mk3D#dp*dfW(wR
z?wRv^=g^XLK03vfw6~&K#wrpyt{5J;dJ)}6nY`C@BujmLLh!yj4~y0bx$6c_cw@6R
z>$o@)d_At?u^o<VZ{BK%+8W7y%8(%OAtyO;3j&d|N3x*Mb2Pe$;d_5|BCc+OZ|=UP
zoo|DQ<^&D0q_u>K?hAm5st_zR8^J1>0kc~%kGaheVc)m&9LebtOw`R4OQy|aZu<;a
zp}H|x^gc%CXf0+|!8>e~+{j}Gc@p&|nwX_D;6~Rw81gU?op!EcGBbm~vaXIhVLOwH
zn8x=phP=@{avbcH7Xfb#MRs-KE^7MtyH(s(cUDQy;H?!Vj9xv6TB`tqB$9}u=ScE%
z*K5uzR+6hU;Mkuq9nx!^hjVQ+u#&rgEj+VrW9l#_e6FGU+K&*Qk|F%|z!EN;$OE5=
zmr&vEO=x;9#Vlvl;LY@P<ompTsFoTDi=sb*#=N=6O-YA}Cypeq;RU+9bY(8Pr!yC+
zI#itA&-pbKfkv1j`#9Joh<tyJQw!udhHq0KY@!i8*FBxwS4f0ev<39^uVY)p04l|$
zW4~$>I2RY;0H?rC_*xLl=O}z!6;DkTo<=zlDHf3Z2Gc7ZbH8s$L28s8Iau69bzVM1
z3CA$1H|{uwYswSJhtpa2hKb;Hehz<DP@_KQT}a)G5~@D-G>S}a#_+)#I4e6APIZ09
zu-U2D$LH6&cjdv|T^`^++LTk6zmdGiSxxq>6=6eO%1j$ik-V4tghrNQ(D70+ygL3J
z^D1-E_Ejo|XKiIO&QD@p9%ih`TZot1054|OfO>-_EK!qUi4}ja>6Iy{q<3JHgcfao
zu?K@y{D_u|CQ8>ofyGr<(cV}S*1Vn0n0h%{hU+oy+GDu&q%N)*H-ae|ze6>Nv%F{T
z6Lr~_OPn+vVN-Z9_JmTrSuDa-?$|((*;ss@ca*%}gz%!Dq4uh|!rtTZ=rccA7;e=M
z1>eodvff`<ci&WK-pJ=wM*ifk^zA|2cPgy!)OY-1=Zg2WZezNL8rcz`N%oz~MCCtO
zWZw2m;Qe!e3)YoK>N1|{wb+xnC-^LVc^-t#6JyTm+mP%(1rAF}N%NWy__)-7{V}<Y
zVk*_>Sfz<CCm7NTQ;NvdnK|Up{;w#=<ag2sg+Q!$SAzRF?2em<v&UtT{96??WY%WR
z)<v8|-#d+$-FjeW*kO`6bcnu-Tnvh$S1=-?g|hxN5V7eSjTNoNX~!dQv40%~eZCBB
zEtlx<patt0UX3>fZt(l6BqICKfr+ea#wj-T?0v;Z7FAd|+eg9;k4+K>l{7`x>Af2L
z#d)sa<{Y&Bo=#Uh;C(28)i@l>vB|$3A=5>c<aDHSrhFd7XGt|RlXr!l`)2HN<RUUw
zsQ}_tBZ>0QN7VX86hvRt!q36t#LuT5-}k*ogTR;g+`k4y!s2H?H<-;5Z4>aZcPjoi
zd4O+Q({OC%Q`(lcgQ(2=53)DShQKwF)YZ2{s5>!+N)<QbD%p{EwRQkSbdCy>c_!(@
zwh>HbQ4G~uw*wc;O=6~TA|!P77OXC9z{)}^cJ4zA*)~Ig>E&-B(<)Wa*<}GU{H%@T
z9;4Crvm7};(T(@sd_jN9a9X?PEFL>>4pW_uVEIH9X07y7_$Nq><t^(IL~PqaO@aqN
zIb{|)PmM*(MUy$nnrIa1+lSh7%s7cB=h6Fd68atK5k}c9qBi@aq35v|?-nlOPOg~A
z+U|ekZiKC-;l0Q4jOQSD`S}vlljGUHd?QAOSAgN1e<+vmMR>hIgq`@&1HKENaaTfh
zSiaRis<Nk@e)#zp3U=v`t<vIbgFN5y`tus!&v%B7hsE5=y%I#X*%scmT|)aXHQ4O2
z4}!+;!uH}?n4ZGVfxK^l?wU{3^cv6P42|acJ|2Mgm;r-x-AT8n7Asn%2*DAR!v4Ad
z;`VC}oGCeg(r$X#s@adu{w3UQ%@>%y@Faeg?0~Wfr*O8&5|$7b#N2a#V%MkrSa2)_
zjel=J<-?gMetIe@pPK~s#`}qJgfjEsj$xF-CZPzQ)&I6#m3&)02_o9QL$D*y(f+!S
zX_1qh>)xM&5zdoY;q}F6#CyBiHlO9^*f(j<N-b6x{ShypEZ|z=_slM;?-Hua7*B@E
zI-x+SjqeF>=T<maAUm$k;;h3UZ^S7<^f`|Ct-pbq@mYAlWg#iqI*iQ)9J~IX4!LZl
z25v_sVAZ`Ea!g*5T^p?oCN_H@BCP?N?q8t3^P^CrDUyuz`GE`4KY_L81>xgf8%Wdo
z4>B+R!B~wn{Hrz!9e$4l>ybCG?vNB4bwGnTC-0}G1{cBltQMze<-vS3pK;o6c(+>S
zM^rq2g3nw;L(#n1nAkXhg#4R@5A!py=e#y~M;{{wKSKQo4e;9$XEh|>BFO!{8Y)Is
zV#zpjmLI7~d@lT=nd!-JOwJ3apcY%+nsR%_JfR{V-tn%%SXy)FEe^bir`pmh2-)aK
zX8#v~4@HyFep3PNNU0(Yw+CsAvJ7-?K$;(E%1zpmOeBwnqdPx8nb2E_pKV98?RReC
zot-8uO0o>Y{&<t1aeOxL{!Es*$dom0HHY`B4zpPu&Cv2w68wUmK}_j$`dCGiRftwX
z*1ZUH4t4^C=wYs`@F(ijNU<>IVQP_<4T^U;7BR4xX7npEwNzDBG5nFrm>uB4J`8Yc
zo}^<(SRs`zdV!wD3(+g=4K&vGqDfd9dQ`ZeUyYPNz2+xMxXj~L&rRal*EgZO;Rw;u
z`;5QW3OQ!K8SRf`V}@oHh#Z|t4j;E-jSt7;+NRlD#0E3&m&sl*$l)2*#Tu9@^h4>h
z9sGOz3l*C<1Ac%13%5$kp!?Ana$tczbA5kVAjr~X&X4QS{fQ(<`Ax?5bziXi-7SH+
z;a?m|Jd0j?k8qk_pP*KQICynL3B%go;#A<kHzJ$&dxnrxQIGLaiv;^QcnaQ}5hcMp
zR9V}sue5YN&tDuh7FX+tp>mKf_iWn`hUX=7qWsx0yEjsJY_AM@4%|dHEeR+~*vM?-
z(m4Oo{U9Y`fQ=1$*wFcz&bL=!GU6vt_VZ`l_Hhu~ZJmhL#8S)~CC!3=+M`TB36zeX
z1{Zn9-J!WRa7O@tR>){UyMBM7z3ZyL^%5sA-=5EHn^28;BVvg5@E*K(a1K3X^9qC9
zDse-SB6J7kQtO|~1$mKO_|<q6x3=>G9MrvqMGYs=KJGMG<9iuhZ(gU)7uJ)B(#fR5
zFatlOpTXdWTA{2~BBngkB5i|RV0%!9c$r=19?eih&8k0G6S@tnE*`>9R?jhWOd|J4
zTM=trX5qJ7Ulx393deG9(<NIL<JF0eaP9ijL@UJ(ob$$$*Zg_RKQ$Si9h}ItM(@P)
z6MLwfmIkY}RV6uJVu+&Va|{jR@75yEY1-sSyxjF1`&U^IvDi)&i|j@@{RmXqH6DC<
zhh@R?rzq`Mj2mArAo2X(Ye;-KXFa-Cn0wHZ-0!RuUQmm{PSZ&sQ4|Z|F9xxBZ$5{8
zQMfvHHMnwLg&Jog@zwnbPVR*w$oW^Hrj-oc)6j{YJ_c-F)@8_E?nyqk-ltVOm$9Ys
zsqo6sH!S%i%Orxv!s;tOVPgL!T>j31Tz0X<yIOh>k?BQB*ISZv+59=vauMr_TnW3(
zgy0Zq%X3k7Q-#$X!ZS{eKn+sKES=BjK1+}IJ?%#)i3RAhSDwVZX@m8Km#DdC2F;Or
zhI6>tXzR39xL2EFnfc-vb0D8)-fX9evZL5aKEpL|VjsEGEk~vl&n7)%Zo&LOCD!$Y
zpNn5sA(ppm@wVJcyfM0iwiI83I+OPh7(9pVIz5H;OGJ@PDSXD}$bZ=4a)thnqB9M|
zs_VkADMXnvg^*+>i44zKJ4KQtuO<yj^DIdv2}v?1Axe@Y389iad+j7isYohKNF_;<
zB2D_v_qSgvXP>jzy6@{6UJygF<zLaA++m#5pw3FvJ_;*!r@)ev&6tb^B<p$+zH?hi
zbXVNtEWbM8a<7v_YyT+LJ3AJySzIIf=H*;|^EPUfQ_Qu?`wLjuIZi354f|%l6Npr)
zUN42|tg-I{4Noi=b{#UqL(g@|SWA8Kx;~URwipBVW)$J9&I@}|i?Hy?91?BagY5<0
zIBY0}f_vLB>3$|RTl_wJ{3c+Vy{0kqrDjC!qaT!b5vGc>F-^7@m+wrZM9d8(_IA*F
z4s-Cnj~pxhEJP9cM*E>Pd62T#pA7r{#LJ&GnB=JnO!;HPT!lZmI=i`Kfx{W%nD<m*
zBtDhQEVe~0o{MUkHkagi|HY!)#oUY;b>Kv7;q-=V)Z8_VhVDVCF!eZYReFF!e1Gg<
z%u7txE2bL7gS7ecXIP|`!l`>K#XH;ILW-&lOZ@N&D=t6ds=L<H*YnSyoBlt1rCvl^
zx({<27AnN?%>lF?<&TZiv>i4c*o;YT*DHsw<-tcK8CKO@ZT~>(367I-z(clC_*HBy
zbL4$yK3W&aI4MHIV{9PpM<e`NDMs3j&EQ$yWR^I6B+1s>Qdz!CoU~T%q3dT&fnBR7
zvOUXwLCuF>IOmcCBxo73?CWB5xNadk<@p6uG<OhZ&oR*LF^%<|JI8zI%Y_TiOR|F9
zH^4mCkd0WSNG=^ngW})z%<^KPV0VTpIoKkFFTO_ObHx&F%j^u?aExR6x8@LuC)v=x
zq8{t`{YTvdBY~;&1*{H?A#*N#fz=CFv3<N>xWlU!(_^NQ#`8bvuc8&?)x5VTswc`U
zrt<8V#HGT%xk_a2Bq3HvXoEtwt6=tywfO6s4PKc%4<Fm-V(pcB=;3F&Dyu!wsOqjz
zp-K@S&L7KXpL4+5<g*~o+mam=zt8n$9YbNlLG+6;XX;NrK?>jTEsYjucjPvZE~#cN
z#(Ezf44F<Qj@yK`v0CIAKM&;;&w}XdB``iMp45#{fyD97cz@Yxw448xyYF3zHN$0i
zP#f5OQ%&~D@D^(D9G$JXk1^qzG;4KQ%1BBJ^o<`+*M2*QD$@hV{eu?7?QtgLIn4uQ
zT1T0?2U^Tsjv*dAv-R`_e0nQ}$mN~Hv-c9I@8v0Y=-7Dn@#<NqKAA)wuTSH??~BBr
z4IErCkYMH+rPx^FPCX?f(CNcw?DBpqxbScXX?-b5q7C0*-lK0Y<lTddO~Q#ng&|of
z9fIeB50cA~UhwXp4fubb%|0h?CCRHd;ky-lCeo+^_sdI?`>twCbr#>DUiDE}GA)_f
zjvUX5dJYh`$Rk+q<iHaB+OQvZriw%p9x6=Y`zms5L6A1f_-V>?WzES|-f<E7cpF~P
z%%THXN=)wHT^xE+k5)zw<hpGPsa$#soj07MK^{N2K%WY}&u|$9bLSF=$3IYGtTQo4
zQ(|69YP<(Ymdr2HV7cLDOzY?>^j)Av^LcJzXJ!;eo)qKuZOz859mm1#`!D>#yRO2<
zA0@x<7GuwFH0iH>0t2K?I3d;;T~B4f{!j0)DxJ`lko~lLUoHA`&gA>!Y%HF0g;o@V
z3Zv4mA+1n?HHsNHw84m}PXC8z0%AaCo+hSCiL>y9Gr9PEdE6d&fq50rX#4JbJXqDh
zd&uhO>`$V^Qh6<Xr{qNJb{4^rXBFUk-;J#}wiG4$OTjJPnk8{6r2BCjM7^5}=a$G}
zUvd|<kS&Ee)5~}_>K9&!--ni|Ke=aB6WFn+25RfN0-}E<3Id8WSgrgQ!0emwMu*Sn
z6^pR*9XHU>YzdLSeU0{O#o+=|Sypy&FR6Kcikqr_34Gf>2}8|qgZGSHuJ`60WF<D_
zdVag$h^q_9bN&JyCyQYSJc-DhjXWDC1PaD$v4mHAmfSN03k<iAe_PH|kC4seu-q-Y
zyHgwY$>o#bR|9Arvl7p?R^q#5Y4EOHg<3rON?Q-?;D#qElCrcL+<>|*yLYIZcev`1
z(nYT1#05{>>m$wbkDtVje+8V3br>nnenBVYoq$a1ELd>w8A>TO;oOeNtYFD;^oZwR
zl|d?P%@8H;X8gprvoGSzh7n{`FwY0|QX*0bj-+mq7?*n1oM#%lP<kK-%)DY~o&9I(
z@O%?yw2y#h^PBjyqXbWW^P`dG`J70$8vm>q;PqYN%y#h}m{_xud>>&g*nPSPsUGiI
zuz87gf4w;x_W(N=^>L<$dU;QF45!&RiseLlVs48(bD6RW5(0mtgv>@(`Fa-Lt6hzM
z!{_1AI1e2A<T8<-c7>{aNWsvaYn9wW4^HV`Bd$6(8|U@kp$mNF&}{hvZl>Y|@SWc(
ztj+4AjXvFi8SFj={20S5J%<IqS|v%rgSWKhy*k@p11M6j0bRfUVzTHe7`r7OE8e_?
zy1(XZZOSQuPrV6xa0aX>%@mcdD{+O5F<gK833#-U=MfLi1pUAwxN|xdmDKom;@pqm
z=<<Zy>|p`|i<4+(e4X?Bc8kj~>jM^V%JiIlaEwtj<_#XlrsiKL#dlGZa)PjT@eu9s
zSHbt9x@<7+1=p8TMRWL_s$;kalU~a2#FJlus-+2^c^QUhowM26qw4JTs&P1UFNt1p
zn8WRRqeQIMaL~TsG(HtogpE&@(d)-t_;a8Zi<f)if~}kJ7`(yYENP-Pvjxlcj%N*V
zHmosa67wE)RM`0XFt)f!0%u)}=Q1P-!&BJZcm^v^TY>kD**JptWV$;?L(Rc5u5Dc*
zcB_qIwKHz>yz&N&F71UD9y!>?vw~>tPk8@u6w~64V%vswaIjI91<@1euy70dfBeNs
z{V2z>^ftVtFoA8;F#x@%6UpZ4hdfI|fqh>cMMRER;kodUq&RO43!dqZ_ct*NsTUR2
z%{eJ-yPbm~n+ie1Bm&K5<w4UW0}%FY#iw$qv@q8iyvsL`TG{njQxZvMTE)ZC3{&D9
zIE7{R57S%WU(mB@BsA@e0ui$!NZzPKR(J9~m65j0Y|{)hata2y)eFgPn>H?oDX<d0
zGuW+VLpEjc9=JvcDz{}Js|k6Dfgkw0qYHmu9Ww_5XPcvn)H}E|#-3R`%*5`?UFe^X
zK^L@%6P+tBG3$C3{_>P&`iX;r@5>)?%VG^k1+3vN42z=H69pC<vWaYbXiH>lv+;;d
z1gVkErh76+K>MZ<XrCX9|8?_BaJ6qjYY=CD*KQzrTm`mmIs&4%6tOF@8l=3=a|NQo
zDC#LfN~D_Tu(Kg^*=|HEpYG#3CDXAXEDB$2aH5CA#fcn06V`q{n`uRi#SVV%Yd+Z#
z?R3-xZ+*nbqs8jX@!v+Qa_r$m$6tY=;#b`8^>N7Qs?hMod%6A@Mx@X0jli#53mPrg
z5s6=M^vAeIkm&dt&PxTdUE7c0-(#92<?tF<;9Nm7AN}EU?##nS(h@97{Rxq2&mcM#
z_HcZe1H1mt3ZxSAvFFc7`o685cC45}>=G9PA#;hu8C|>+ejI-*Pi3Rama{IGr&M`l
z4yVk|A7`t&bLqU_GQcwt);RnXbhf;K4@;kN&j+@^45cGb;?Yd*RqtT^U#;0|mxV-}
z&xGpyH6i~dyu`xyM`5V46|((L3mkXx-iozTN$#-(>T&TC{!%#!>A&U(E1$jr?cQrh
zqK5Ex@=<bJn1|1T<}vxW`8Xp<gn5rXMM_svO0V7$I+f36f0M=8p9k$A<GPq7x&@>2
zCo{IY`3Y?4e}{ofyzxWMHkRAlkN@g3urlZ~G`p|I$*bPM#kUjLH`NoU51qK;!&YdV
zR?Ia|9Lp}Ak|o(<;^h38G0d?_g*+TPK+E1%LGtY-_$6PO#Lm*ite1yCQZg9)7NlT}
z$3HxI6Ij5{e9n4U8qXNqL5r9PWba{~-IFj#br$o?@e`w1)PXf@cy$n%_V7KYRiE)j
z`~vnP*%54$y5X7#;DH?nNY0)FP|Gi(cB4nLd!e79?f7Ht5lq2ySvA(ADv#YiKG38G
zX@Y2;5gp_s3MZYPBatu1W|hz2Dti&4eqZKyl7*P~`wskV&qcH8O6<*<5aM|`nv>rB
zLhwD`oEdwUa^D}>a*bsMZ1}De`#po7H`SL>dn;X%m!i(RCpdy;Spzk{RfIp!PiLyJ
zgAnfE!(Ff)O>|_|!Ra%3C|~gzv#)Lwx=uA@imPOB<*jqj9^zQpC*9^?6MCAv{_Gk`
zZ<Zu~j9OugPb+s@@DhCf=E2+>`RG@cK<;1jflBWLPS?DYqfZ;Tw?qoG+oJ{Eg1y{~
z8$VFwd_v`(<y9D!aErTYZ@{LnT1G71e#Rr02jSjG12p!PV-Z_;*8R4>@HjRVJ8G3!
z`k3FGB>!#P5bRIBpC=e_L5ujr*J9zdY`UPMj}x2r5}g-~WPL*u$r}4h81MRwh6e{Z
zOkNoWbxTT!l?uS$ni6OkQpIHd!#tD2PY~7L!(FY7MD4?4nEm3{)ctWZ?c&yhW!`!8
zm)?pW*4i--EQQujYjD@(8{qz4nYs2XBBT9AknE)&h1rgW>4YsiQEG<)Xcdei&wkwn
z->Fs@ZgM~n9sY;2OZrB?Y|$c>KYyS^=~Xb*(_*HT{lfMsZv|=dp2H}+8tlAk$*zu#
zAb-1BF-v%fcOOfV(59)hpZ)~tl}T9kU4tdXgb8hwM-%<Bt9asp92?622NwpiVQ_Xe
z2_Hy?P;#6$B`A|ubQmX$(PFLV|H3`1Hq?E836tK`2up}K_kE!_pNCVy>r&StcJ~xE
z<TaLvpOIy{pBlOFya~jp+JeYc#}d0YJ4oHIK$v7|&m)Mma95`jnZ0fuwz&I4+=UGA
zbo)-H$#)AJ*NL)=N)4Dj_!QDyJ_@~Cm!rX=1Z@7Aj(+_gQMGjfTmSAeHNBfn`V*$X
zhWY=&{TtmhEiD#GE+=s3t}C)4pAm4tMVaLV9D<g~lEOgVSEn7BgJtxG&_mM`Wu9bW
z<kK3gc8(;?JL<71kHA!o%b?(13l{bgZ1#*m@amsV<R3<3V<SbQNCq1FjF?e6?~zFr
za7X+aP*g$!PP#|nh8KJ{|IKS0Zr8w)yhK5Vks%yPFvSDT$*5Rag`MToNmaTk4RID@
z>qMtBGo{m1ar<Aau&EU$Ze4|n>&M_-o5vV9?i1JIw2T>^si3E=lSy&xJ$$O~Op<n=
z#O^an<f*PJ%#-AE*+yQKMc<-eyU-6?UXFsm!3KOBEkYKZJ3?-K=H2o0^=M$?esr4f
zfKGq;95)?5#h+z!Fh#=?{gcE5a;8&&%a5a9^I|c6R2bJbgRl=jwAq%mlbN-8-v9p$
zf~G&mm*cB&|9_MC{5X*QKdUioq!>y@O(N5>HCfrY<M?Fs4JcVELoGx@q3h>m@-x2!
z9V-JXkBRD%<ohGYo^DxK*RmB`YWefBGZRf64&XC>*U&z)2W2)L#gL3w7|`^Pi*Jp9
z-kXwSNiom$er}2e8%}|k+EpyvrbU#zoZxjz2G6AXLD!a=vJ2t-8S>;WzL*dT^mPDL
z`21DieMbNqag#}V^bC5X@g3(Ja-1&6QKRm0C-GTW9-MwC&8nyJzOR~oIw@EM|Jw9{
zyXI)Rw=f6fk37NovO6)H{}!v+T+Q{KQDcF{3vm3+&zQdK2{&@_Tg)lq*-WZiA^UW$
zuwMNLMvnatdR~u2<=xAOc`xEUrw?4-;|aVk)rVCSh%hnT#Zc0t3KjCIbRkt@dtaoW
z@;)K`aNYr{2IQF(bb<bp&sc4~iDh0Jz$DFQuwJDI<vZ-)>~&+@ANC7&T-Rp-UpJ8G
zfvIe><$D}BXh1@&$FiSGC2`K75zHj8j0R2Y6NXu5LdyAc>MeE(16N++zP9UgqsEq#
z(Z8K>lJ;m4JSmY_=4r9^j(^-jzMH&kgDJ^cV?@SG3&l!BXSP)=g(Sz$WAPVNi2u+@
zXp2cA+~hFWUOa_!%V~noH_e#Wu{Jn;YX_UA@gIvbl>*7Oi@3pClS;%)U_rl{1@XG3
zEFh@~%Vg@%?BWK}cd!|)<4lOSbU$*^JLw<Uy8<yK{#$8&5dJ%O4~-j*`MdL4a^kH$
z`Fc8lG}>ix>Hpf{LC{1jJFg7E&SvO)_AthJ>$7x=HncXqgQdY==!(@mBl@&Hxj0-;
zdsbWEM8@Bt-FL%^5yyl!J2O!&%b3hr(}~({lTe{k2_<v+&T5q-^PJs>fm5Hs(-mP@
zmvWqIkz0j}ba=1A79D^E0pw=r5llVe0G@-{C{lS7?Xvef%vod#>gOxD?4JE(ukR47
zNZ|SSqsEhnr{P$hD9IdNUgxw-`ml21c>Z_uFW24}g0G@h^Std^2Ww)Ex@L0Zgoh?^
zvk-@;L>K=q9#7JwVqyQ*ctKBi9Zj6R0}g*Uh0n4&Fmtpf%W`c%o1~jG=YB1OTE3)B
zp*vvcZ#VsDqRZ^n4xrYpBiszxjnE$;4yVjy$h6S8Om1r+F0cFs;-9Q=d&W`J-DpL0
zdk&G%!+dUdS^}3e%B}LC>p8frz5~L>?W5&~CBWgyMNH;%m@;E0kpP>utn<z+(yCp_
zt=LwB!i9%OLDPQnx%VDO{-=QfC4YGqd^)-*yvAl(H@NuV0fvv!BFYclX;n0Ww?Kj!
zl-6V6!)7QR%!A~LZZy%#BPs69f{3k4$fkq(Y}B<iJiBck@i_9GPF%bU`t8h^xUnot
zRWu_ly(5_WrVjdj&;}g~b5OWg2c1<E*@AL4d^D~W|J#*?x2C_uCgU@7pMwM(ejv)!
zA_&>0=g#V$ZzOMnRfx-rX=JCgG{&!eg9lp%=@M~0Hu<S8^Izi3z1&p5C6q}sQ==pd
zZ<HcoHoiD!W&#bLE=ugCco73paVExRX7um>M;f0mK`+@bI)8;PE|Tp)gC|w!(6oi<
zdQ2tr4p!shJ@$|tSs<9Oe+fJ8=#NVN5>UQk3pslC9?lJ!%FaJ~g%`JaVNQl0HuD|F
znU01`=`7EH+Wv?C{=S6!-y5@Oj!RfkVm6k!OlJ}QZ6jYs|A*O<=lMLq5!}7w3f!2s
z7ur=1ID{vaa3fdDLI2@(!rW;p@a4u7jIt7Ef$67k%Q$)XGcgrZH))}PuRe^OCC2`w
z-o(A4!+6!*9v|uO=jH7%(lq=Ud{+m++7~?YLvbET4_crUv0#@2>gb_jJ_p*z_htV~
z!9w?b?!Npi(mr7{`99hP7jISI-OA&H5<WF3*|Z<6E2a2;e~D02TZR3%KLz#9IWUJv
zZ#GcW1TNu6A^YTH!NcrO%AzG$`Ti~_`KU?EZ4|+1dmS1{Jf{feZ0c%Rs1M`0j&F;(
ziSCj_#A^w+=#W2NwBfTNb$4iV@-F^9U_i{4*^=)Q?m$>m6a9W!gOxTW!!@U1vV&D&
z=2t$m(jH5T-}2et<1*a1eOF=A!#3{0S1nTek>8=1#*)TKhJurRQmirdDb={3#mYZr
zpzm=_Vcp_rZh~|PTq%12={LX9tJk$jVf{Y*^;n4vEq;R)1JU^Hnl~DRZiD%+)$x*d
z44#|+3JR(>;`3XgZ26iXls@|kT|xthp5qhj8GD}6OMF)*VFQMed4vUYP@^~t<~~`K
zndyjfn<Hj&yW46w?VOJ&u_y`bj^4t$bbYeu$UA8NQ7C-A<|vq^^1GulXR>rzG8)}V
zA<fw-B&VSfu0IoE)A(*7*{FcqH;rUnI=S4JSb5eLR7mfey3jj043l}6&qC9&tV@43
z(Rg#6Tl8o==&UQp_iYPVyUTci!m0(r8Lke@GG3PFL8`I()Y}*`a}=lZ_%%MCBS(sT
zJK^HQaL}sMf*iqd4Ey#O*Uo#+&E0+tj(uB=5h|i&?K^W^Cz(PP86APOD_2m%HDh7X
zL^H_Hv!~gG2@V@x*Q3}sdG;~lC8|HG!;VLn@K~=1OLdJQQ*WNbp&Vsea;F!op6;)-
z3lE^ZI>$kG++B2#Sqc+o6>#zwD(S@Jd-!X^YW`bckQ?@SBfPTVHa&NBHe;uE(z1WY
zVaPd0m|D3VpX9B>wMKWi@Q4D3(fxe4VdQ(tS6exw1wn8jTO70a?2|~L=k>u<b0)1G
zOpWgChXEfOp38iL2Ig)cLqXeVvG5OQKlj2-UOlu+KLxzs8&Lg?fAH#x(d@?}Q?}KC
z=j`R?0PXU|VE*62!sswv`;K?X>{euz1}BN@{f)55<SjnIx7he}IVqYmiFvw>!|<Of
z4ptwPnTfO>4EFDUh&X9xDDA^@0eZmGUz!+^XeyPN%J;B7g32epyYa%BEZ62|dYh-i
z#&h>+REGko2=Ju$U5ilvr7TX5=XrU>QY3EAA5;#=29Vdr6~7tgYNe9x3wd6`$a7@J
z`yqPf?Od?vFX383n`q}#d7}6#4sXxYARR}9@LF+*tLQ%~I9785TM~!SR_`K4?mtY*
z>i^LRW2R%;)RVY$eI&N&s-nn<+4LUY!F;YZ7dl4v;uXne4A=Y4h5HmkbU->ZPudT`
zOJ=~6yS!UPZ(e0Q&*s&>6p!??B$}D`)63)2X^Vphc`YkJel&`bi~<?vI@*t%eZ3A{
zQo9lO+(&@)Ts9Q0Nr~M7s<ASk3z+^>*nMyoS#6aG;cpe0Xz2`GGVTzcQ4ob>JxS6>
z{|Wcs|I2d%o<rT_E<6~o#PkaKL3z$^Tps$Fil`oQFytBczFFbqm4ghM*VBTpgNHF_
zB4KWG`TK9PEl$xGPt2}p;)2@ef+_3Aa!qcZ!Q@;x+49AR%+nMh*Q*oB+r!JTW4|%U
zss7Blk3UV-4o`!qQ+Lray9JvzyOOIR6PUBkGtAmAO1Kb4U+9aHcumCZgW}}o>2S7j
z)?aiJPse*pAL5BrJ2Kbk6j^O+N<x!TFygo+i~A4JYEv$4citkr`_&WI1e<YdR$YNH
z`U{!Yx@Z~}=*UWT{>Br_reg2l9I}A-zig5FjX9M|C{=BxY@!Dw<7Y^(-v_hZjIkv+
zkp`Gcu=zJqXu9nfdN(W&-^rxW1sPr3nqPWkrsg@U8;XZ7Pb`?EybG;reGFet>EVm9
zuR&o$JLhe<k&Nnli|Ov}SX0!5ks%%Q#n5!PaH3Lp@98-5-+X7>y&)f^_!(!c$8o5B
zp3cP==<rM&HE?{rPdG<Uf*81rB;Mo1>BUenT;;WooNW@q+{dcKyR?vtaLLCc!^fzX
zun_-Q&m=cga%dMnV^^2>%WY1iFuv|2?sndVa-l(>W&e&Q4O?=1c%J{7J~@`#c?c82
z-+{uKG(pzL0Q?}g9$y@4#-Cd>@u1;IvgyxnEIO0}?TWXsMl>2_M;YK#=R4Ra^GgtY
zIFn=~M`E>0BPM&~(tH&^BCa?QJ4Y@gK{1i!`HXoiQO21x@_s>c&(ZAn_E3y-egv*z
z{C|+9E;HtT#(i`0VB9@D5;N%pDox>M+@|u78T$rjhs;IM_vvK$G%c=F;Te?5mf=@^
z&MG(X21oPGMpxbqe8b6z85U<^$&BgbUGr0nw=2b!$A-{%%NMNaO$Gy3QBvW%fmlv^
z%)QH559EUt@y;DXZ;)}U&nd|vX<P=a<#*W;YM%wB8$7X^KMz+w<~jJ5=}<{7<8D49
z=J8@Xt`irKp*dTyX4_|6etrV;)=9!+zteE%v`)^(Z8TeAcoU>~-dO543pP0P82lBz
zA*3iD)_<ReYfCkG7HT`1-s#7ZV-o1G#Fnh5QJ68?fn_V7v9CH)j57D$BJJ4CwfhXf
zi?}&(`fW7tRk^{lD<@$`YBcYnd?ILX@a3Y89fI`WyO77<1xF3=e6JU25WlDx7wqT1
zgTz0h(~~SNar_a~xZHp;{`ug$d@3{QPesqIy3AwX4>s;=<h<Js@Z5?HxH`>%@85rg
zsv9Ta#@PEjcjdjn^7FY$>EG>?%)E_*Kbk=KeH=M^`X8+CoJ;acp7WlWD0DSx#g>r|
zp><3FN(M*ZANfTv{cj?be*O|$6qeC~kk44<s?N$a$Fb@cqRdy(mnQRB-u+E~&}8uj
zsQR^>E=6g!=+afZc`1U17#)Rj2Ng^{CCfZU@H^hU5oGshTQ)S5#pmeX;GauNS@t7M
z6ze%pLqo;c;jS2P7d*smibg~u<`3HW`4OYK^O)Y|h4t;;@T>L?9SF2%`|pis31ja;
zv)(TZwkd<OzGI+}o+6lFcL)zuY{0<OXV`!35^Uy<P|=Dga9{j}W?MxGn@eB9-<&Hr
z_smQBD@++O#_*lLn*-eC1Ns=$=!iK%x+t-I43Ve|;Vh47kf-uF)XP$Xc%O15wIRD9
zWWEJj?KeT^F{dypMVw8Yd<M4->|}NNV_5svNW2xggNXDU6+ZAw#+xQmB*k0}P3(|r
zFA!m)eHEG9#JT)^cPE4lKA=nIex~d~2c0$57q6cIX8BzQT@RjwfT~w`C~YI=zT>m8
zl?!qDJ7+d@_9MD}ktDtoq{t4}OW1$u4aN_ofciiMr)8T7-D9no<+jnJW}hGF<V_EK
z-$hs&zdtYva3&Hm&%p^s!|DN9R($mdY&&%dCRrGfM~>1=!)Q8LAQesZ){lp{{q5NE
zVI4VDn$9K8;<<8@PGU^x2TZap7Z$7&VS61jh&+D}_ZV7+Y28<;+ix@I9H|47uBkDV
znPOyhg&q_ex8lRB`EY8}4C*^+GfP;JiU!(kF!gynv0H1(8r7x?wL`{Y_MQEL?71?y
z=%0{hsTK%4%zJ<<jYMxXeQF(+Pt%)cu#R=L*!S`k^|bX7c17L;8Lf?YE1(rcjE)G;
zpHRkyHV44Qhi45hvW3dRG;H**MBfwcg7j;a+^)HO5Y{^t11y@beeN}3^UZWj%H(tF
z9s9{BVFTV7f0k(VM39qTrEu-nZN%>LCGM&9NK$8)LFMwVqUzQon1AjDCap~5T;7M_
z$)F!tI4>Ffdj8Sy#}%A7oeo)xN0AqsBtfJ3BUXf4kcy8nwEz56G`n*JT&~+N-xO7*
zJbe<4pOec;nSDpeoY(O7`8p5@+*SGAwiRRFim>OCe?#8iCfxLeXX%c05eyI5laQnQ
z{_To6dnuX#B^i^+M&VQHqjH>#c{7c9?Ja_K`ztW`HxhNf_f_t^;mz7ITe;J+1CXF9
zK?H$!(Pdg9Y6ZwZ-_%{iPEwwd*TA;NNHJxXF>L?2jZiak4jx{-9{O{~pv2piRODjm
zb=xU($kbyZ<XOBK<}EY8na|(hCY{$1yV(|Nx7@`qQR1xpj5+ZQ4Mq<~d-8Cd2O7BO
zlk#PUaA}SLH69wl6>Utwpg+?%{}o$kMSdT=OSwUJi?%{Xl^Mug7e%gpHH4N=VfSyE
zlEhp5d|pu&jC|`M?Zrxxdc~6%y;#Te=nMSVWJnZF9EM7ZM4VdKOWm|QAS0P)%<`Ra
zZ=(ft<4TS`e5}q)T=#IXj|}kmawRxoxe~s^L!7K+#<brwQq58E_`qBj{YKAaXY3ug
z)&0uE^JcPO>xpMD$wUh(Ux#Ch2%n2JtP!r^v#a{+)mX|aeYSg>EmO)K!dl*!64AJj
zWc??O7h-ts>9K{ZNbfA9@%J5@E8|$RR4h*0IE1s~#hG!;KN?hekZyiiFDO)84lWN>
z&`~5`sP~dXL!NtnG}DebgdHG$I}>pAJ4N{VABP28EcbG>Dg;SPBD;%Psd&#9&dq5i
zyTwkywyT+NP9TaK*T1V=+;@>=ok-(GKkNgSaSzaMNfV5jUkJ0id{EkS07haVR4$Hy
znNjJ4d;gE#n0c25Oc>8bCIyf!MzU=8?|Iyo&PKc;s>Q5+p2A#5Lu!}*5UHUtd9iyZ
zh<xc1xS8{;(R1lAp>7w<kQ>2*;<BjgF}|;QHHM2+D5O=(!_c5@6ZLSpggZhBS-nJw
zy%W1Z<ERzOc`Ad6OH{%0wHscQHDQ@S>0Et46Ap1!ESyf@*|+M%=wuK{d9TE#27Lmj
zS;2UmpS7|+3AWZX8*>M~f_&R;5b4vkcjEI-nmyiZRIUUOGu#Q=wRe%56^p<pEeYQ8
z<h}QWuBhq~iAQW|Alh>w*X+yhs-M2(LgW46K;dF|*F&hoM}9u=-kiLcUWKg*z?#4C
zeWI-^QAXwnF1g@_#wY=`6Kt5+4jJay(dSTia}u*L-h)<`rb6|sr7%y8!lgesctz_O
zr+w@urc0~S`!6)mdQu{_S+Ii~UYZ4Ur9Uy{Tn6?@8L<vcH~gyc7)Ac=#u0pvW0FED
zI;`-)qi27hb#o2$8aIIF!;22Rzt3W6X({*FS%D<}JWOmauf&!0S0MLp3^*;eW%G0w
z;#a59%+ocG-dw$oCT!3kbj295JR4ASwJTfs*qoIe-a;<O7?Q5{%jwoN&e&c#hS9M+
zNA7V1IQ-ZK(!`eBU28^_E^c98Y&fhF4<`q`pF#@HUG;wR6(1Vk1iKU8(dfrC=B_0U
zw)Ow9zI}YJeeP6xbDap$^;H(kuF@e&v=k>DN<mY;!x<SpLvVc53>I(#F)FE^_i7lT
z-S7MODZL6)_t}$(-yS&p;2gR>$>AO?`+@h;RmpVgeAI0bV%m8>6#HueuYPFo+}!1`
zFE@icesCVM4=xslhaSelum_y)xYb<a{!V=I)t(vdljU76&G@~bj_aB0j@Dawe#Hw(
zCTgRD?af6FYme4(54`W9onwp8Gb9f}i#EXQ2p6bu6JuYOokjISx6wJ4XOZ2QNc=mW
zqh6N+%hDRl#&26sIJeV+|4he_p{o7l&Mh@u`8Ng4JwwTBes^qnv;*#r6=yl-X#$bc
z6_s@}=abmI;%vq1ndm(;N$4GCLHt$>;-%HS+zIUpY|o#9XE*cy;&stv)vB?mJgSq6
zmek;WDg6gO{><l{F!mVHI3FXXND=Fsspt~82EP5WV0Pu>2)%Ivjos_0n9?}*k)I#$
zIGX{FyUd|MS(K><hoXMMPmKIx!qQ)TM3D<#+~;dgpg1lc{P^dvCB_2RpC3si<)6XJ
zc}vkek<W(IhLh9h%E>dE!{|Ry&3UZ>8th{Vq7N@&_o`{^&j&tp78nkLg@@?(3voDf
z^@5-=s2UWCDnL&E8|3bKkLIoSvB}U9_h}`d)09!nIq0RpHO>&()@!&$cOpJk_zNoe
zilnxF3zXQnGMl5(m?{Xz)(am|TD?K&7U;#Q7M`k%&Yex94g8sCPc@3TAArrDim<&+
z0xUN-2t*uD3g=fvVL-zjP~hqzFL)2?tLF=rZXAU5E~DU??G;qwT|_kOBAR?LXV$y=
z;n*2@wsDLLEANq^nNGE6P&I`)eyBocy%<Og>4c(<$2m^929z?}>0r<UROzzecl8-y
zIcYLm$Y)Bu*HIdNqChY+JCV#^AH#FXKfs=kqO2iJg&oW1_s4&=;ML@VIBoxBo>^)^
z9G2$df<t?W#Ilnl`^+uO37rD>AIdV5%lZ%<bek%ueMG^GLQL&`1>Mufu(ISBQWi#7
z;O=Gkdvr7mfA8QFD!<{LaAR)8aTQ!tE95e@SEG`*6sxXKg^+G-F8Qblit*uSoZSh_
z_iVtEulT&5eFEvb(oXHqUxpx4-gCV1AU=5S3bzxLaGyywpAVb?rkP2kO<RKu>z>4_
zhXSrr*^*v$T1Uh;*rIi&7ygrvCJIm^OxmH!41FoKZqsBLzQa_%$%Y;9zR&l9W$>Q%
zM6TRUhd3HUq0JvBcJhcNQ}-E1+D&YP@gb#nNH>laEEz(v|BS)$=N6KoT8|TsHepy%
zJeD0B;`yUI2kD#{++S%9!AG`|zB(-&$Z{Z0ryU|E*dp9rEXEvD(}ejY((vu`L=3nT
z!Ff0YV@>QUaB<Usx(CLB)+>kMo$40W7@sSs@G+$k79G@mLjrtlN+qXVwb&#1TG&3Q
z!d@xMvASI=+0I^froH1EK8)P~X7ZxMqc#?e3`fwtN7~S;bP+iuJ_E8&-=<2Z+v)e>
zF9IHAP5h5@wDp%B@BO@mSwSVRb<ANdc!fABM2pgM?#`@LK?<9G{ejJwLrI%;FU&PO
zhvi2jN#s05c3&ZpMjhUWwKiu_Eaf)0R5613{uvZ#F89a$#wg*u4rNl&dYW{OQfHm%
zqez+Z2sWlN3g&J!Mn9Ql#3@ml`N>pqg{c<IQ!jv9&OhhA4+VlZlN4Er-bJ{4O9qu~
z9f(tMF>aJUggf5bk&KNz=VUkEwH!ZzWqvx03)7?EtztUP7rhCdH8pggON-n~Z$ZC$
zdHmDVLLDuZU|)HZ!2d=y<}Mz=Dn^ELZ<MZ3x7VY{qAmq8!#<yEO-v=n%SMp&57z}f
zT5~b`s23&8b@YehED+h8<#6_=K3Yr0;Oae!WI><}U0xU~$i2G>TYBY~Na=Z!s8E10
zr%mZSqbKls=na<CR^q4wAF)eXpEf&H;fK<>Xrwt29W<p`>7`K8`p<-k+*Dv9xwMiA
zN3fD`U+nn1hb$D?jtN#TF*oEir_KqvKw~=+B05NmQ=M6M)B%B$_;Jwdt-(l{Ve0f-
zl*HV=1qF^eY+?LA{4m{|Jt-9j{pl4#!MHnI?36!{qcDkCirO;Odxmt6%Y3$pOvSdb
z{NLkr3uddIB&?{L6Lb@{dE^*q-daO1md9|T$97_e`7$<dOFB9?J>ZW0R3(xs=B)7U
zC}Q$Do4Zim2(QFcxSiF;?ERvtO!oN_C~>KQp^hG$y!0SQJ|9PRoqNSCKIg?}UHJEz
zy9tTTbE0dd;<(=n*K==IAviwz4eyrw;fM_;Xt(7&T+mmAceXRxd5>Tcxx$J{zvB5x
zRd=xC!5H?!BZ(ZIH<OKCqQicyw1<0oCQS4f&&>9f#RlD1cx<Ht`xC7}f+u{zi0^Ua
za3cTu(z7G0cN(+qEzO+1-*mh){0$pqCNT;BO?;QA8V&OvfX_BdlH56jX*2k~YPSv+
zu@`95=)wK!J;JjD%g|MNF1NL76cj1T^ImjE*tAASw5DazXI1qeQM!eQT3Vxi(K$K=
zBghJ42KPIIVZ~z|_IvjzRw$)Px{ojnf5-FoeY}`b(=9MD9YJmy9wb(#hw)=)0zCJA
ziM<-}(0wA3s8%h*I=;IbF@ksMu4|w}E~DWrzjydIw1#_i^gIok^@l2ddqQ8B?ZvHY
z-2`tx@NU8;-d(<KC0@?Fi7)&tNVI+-M+07<q1tKWE<B@6vKdgJF`A}-v_Q9AN@NeU
zz?Wrl+!H=C9`k{J=S1`Ovwaa<P&f*vvo0)Hs>&=j^G*T{GroK99+oQJfzDMkAtjc=
z6_I-wPQHPb(J5GHGJ>67zn4fBAvS*+#VwxlgJv(4VDW1rX+w+^CN`*%k71L*XJ!;8
zu0@=wx{j{Ct-!*sF>Lo*C)lOF88e$0_UJx`9Z_PaFhfJ|<xn<WKY9ZL%`^nLkNG)#
z<SNo|`6jXN%tF21Z0u>2XLRaiw35mLg%PKa`~8wzy-bZoP<~e(_Zk<KZij~nN5SA}
z5xlvfPDZq9v+s?5v|gae9M(L*mrLf7tt;fvxKsn@EY-#Wp0(w;ED}Sn@8<h~BG)BS
z;;8JTGPFKhKxxEovSi5|CVhlws|TKiL@B`LKfSo-9q%RGB*TK7<;mpC&zM+m3tkTz
zxN?X?-Npn#_>B9)QzPtg(ZmuEjY>d?BpdvF(Ts&`xk#GkU4hDmO1xW`%H?gDh6`RL
zf!9%KCX!!at7OrDUeY^pSD7X$Td&S4)GWD>J*P0qcqed&<KR=U0$bmfPc@|?P&z7+
zE2=8MKXpS8t*uISiI|b(x%s$MSCd#jR$vK}wD~OmC467Hk!|)qO457@9ey*Cl~&lX
zpIy34%kme7^ITMae#g9QUL=^U>POX3StfqMm2KGAj17iE*f!xbKH5B=)r|a%>y2yh
zzNQt67%;_$ibpUyb3W!q_>+q(d5?4O4Q#rlj1_MuLsCm1-MVQD$^@ihc*#14i0YZd
z_?R~P8L9@U>N#Z6CFGhX>#%cYQgOi;2{LP&7}UM0p~D(>tlzy4_7z*RZtp~5W1@f#
zW)q+#Fq8JR45P_50~S0Og^o`C!nGBAKBzewO#)8hi!2>7HtZOz8G91{?V7+kVq+m-
zP=?&fi^upwIT&T|5lbwNa57yb^h3Qrp1ibzm{=``%}-^C)+*i;b3ulc+<1(NyY2D1
zA)lN4wVJf*-{x~}sx0xuZx|MPj=Cv*0vCT5Ql)SRJFkqwK?x<+*wT&p!c0&$k0bW0
z5xu@-;_VG;(EEzdzJyt@P6cI_tyOG4=kgs?*S8_tiq?Wlc|3YsWC}f2C1G^T4)n`U
z$MY*<X=fntZosQh^Q{rs+YW45T*+CLpXI^|=V3SvK(pj2%*cSxD_F{~8DZzpLGBWG
zv>LI7Y6-&D)qvu}d>lVn3=|eU77A8X;g7YCpytj~Z2uvkkE~=^Ou`n>_@>5M?Mt|S
znfvjD`*Kjb97Yp%mBOx$=_I>-8F`qXLn0R(hATRQ!mXMaaC)gaB#c-D4O1P++jogn
zV}UxU`#YQD9lizcZ?s{U*-=hbR*c^b+LBDuJs97v&+4kHv0>;L*7tv=w!>9OJBzvB
zZYJD4_g}(Qi_YTIk2<(gL7w?ds>f9peYjTH7!^19gSL=oea@=K6uS^idM81Qd)Dwf
zMRleXHk$2P9}Na(OTZwCW3wH{v89K3?#!{(*!8xZyLn+b`A`~y7Irgm?WzX6w=WpG
zgM;DCo5vtGp5WGV;;bR`JJy9YgYo;5oI7_1%WE5WMo|daFXGOmG#_EBwSX+Q=lLFk
zA`l<hiLp1&g8eNC;?jNw%dUICWx3l>w@4IT<;r3E?@%0i{S(_J0qK$Y0V{@IpsoR<
z9~WQY^d21LpPv+1_eG1XPr1*1t}jO6){$tSQ;Xdl*D*Fhj=8oT!#6qA7@}p1Jr)-*
zdeH&eA2|Tm-S^`vPLr)}t>msW1cL|fe{*tcfvw}^F}&B5c!x%zr%xVH{p<h>`rbiu
zK^SqV-z(I4D^30_&gNE)lVOV1rjRB_@z3!o7>-WRB=Q+??pY#I=W*S;m*b8bkH*NJ
zM%ZP!4Sm(;b7DKp@#y5QwDQ0phU-f)@f>#|;+b!MeMK^6n3ywN={j;hEdkV}%JEwI
zeEf@qB&{`IyY`JlZ`HdP?LG^9x_EA$S24H1=@3R9x8?lJb_=bAb3otWELECxfYY9^
zPcX|-pS(Aj$addO2kkE_LAm=rw|&ijthmdAB+dVc%clyt-pS7FLiIZ;AyWl4MuT+r
z;h8Wwtrz_to)%VA)?nbPRy-5=l`2SyFw2Z&&fu>l7S}aXy}<}fkDJAfb3X>T2je+y
zSrO<x&+~fuJl;Y@AYWq>p)&d&IWk8c&g29_vDGB_xbGq;`5YoQ?-k(eyGD3=uQ}iC
z`GqT|FUCIpekju3Td5EtPX@jCdDuY_;O{dacZ+8vDTSlwnO8X1{seLc*P(u221<yX
zquUSjEI|i*;!!Hb{`rq0R==*}BOt6&EgnRkUBX7eb6Bds0BV*#=j^V8;`i3=FmbCn
zZkv*V>&HFD;k?u674RAVH7Al$ch{4KX~l3*UmB<7h%$Z05$yK|iso0JVbXVLy4>sr
zSbJ1ps6_$3+jfFyeVd?^xHLN4jzIsL?x3ahm*4O3&mB_`{xi(r{9pD7G#x(Dsx$e5
zB@0%NH^&ojwjt7~oEr@-`_9q4i)6V_h;k>-z?`EG$gR3?3@kS!Z|&A$p$yNYkj_GG
z^9fGjuX?5Biu;(x6~LACJeLe#;^$4SWa-&sTt{^b?u{$vlIBk)0ZonEtfV};^U)3J
zS^Gx#v6#Ov^0UP38V|vV>9a{I#KM8oHmvL32%<HU@5p7c%7JSED9R<EZrpYtr!V2Z
z+uy*&a{}pncMat{49VG>hS+2?fragy02wyp$ivlf)MQ)>x+iZZ|7|YBKR&s{@|G5q
z^yZ<0wLPqnY7x{)I)K@!JfT9(YwG#fkGX2tv+9yrEKKA#mHe&`KGLI!cUc`gQsH}w
z1rf*@N06M(sU&`)G9zs^#94GZ$*JNyF}z>g+iW@cww0gltm(#~ASHGpF`rgA>0<t$
zYN$>6htZecqrFllI0ubj)k|9;_%Q{qkM7X1>V@z`eh=E|Paq5XHbMUBf7CHDMUX4g
z#3{e6MYG@hod3;4frZIKYR1nya~F1F=_h-T*ze8Oj-Eq)ZPy`h_)g$Z-AuT*>L6yh
zKgQABk8$6w)wpP=3`@#`iP~NtG9@Dlm%1f$%U+jakB=G)<*D^I{SJ_9r&oe~whwT(
zoj*1!f*@;zHut)q74NN1C9T1`!F+>0Q@OGLI|C15+OjLCHC}<_e@Vv~J09bh3z201
z&YyVb6wg)KwwbK&{YHn5bIiJ<hi=`rly`}mq7R>s@7ePhzw4xe?uYL}Ya=&i`%0dO
zOnzQ@eJ~zP?*72vT9KS`!6<nDwGwrY#M5YTKdzvX!<;h%c%Yi!eLR>$Jot`A=vV{X
zH!~SjkK2=q3lngu*+IUCF~no+GU?#TbwK#u_dv`jyt1K^dmEYxChnf(@bkO)alsc{
z{^>rpFV!OfDwpAhcpPP&s_gn1E9ND(g7_-b3S@mn$e*W4B<ZUMp8Pm~{XL7IbJh~n
zJ#)oj*HwQI=*}R$t0uDO*@492!XoTRZ$$HYMV6Yj6Mbt(ILy3UfR8I?vWnc}_+8Nm
zYJU8Mb$e2YUuQMMz01L|qh8P+omM>m#u05U+{Y)9g?N6mH)QzrV81{EmX06hbNThS
zIpQYmb#Ujor&HJ&#Y!*<I|p{&GR$2;i+Rsd<&yclPIvbP2)eWz4vd>fde>TU3In5A
zaMm_ZxpWldmE+J|aEXgMaEdJ7VoZGd&Vou+22FeYh<B9*;MDm)X~=Lf*O6~R8a7Ab
z?%&HXVE;>aCAouVw#qWa{(A5$Qe(Yiv>~OZ5dU>8V7dHv{p*q!SOtZGfjI|=4pSjj
zw+V3x_Qml|>9EgaA@M%MlUCKg;^UM2yuC>Z*XJJ~X(XGDIW-;gD^6kQcp-eL`Uj?6
z=2-W-5?f9x;m^JCX!Xa9O%-#+#2=|}<GCh!O;JR1Cw27DyUO3KBuE#}>)SQMj#ZAn
zh`IGDKoUP;iMb!Av5$j*mI7|!fBVsjpQ}aKE1;eP@08!C!iGmIf%f>YO25obl-8G|
zTSys7<;_8j?cvz4iuWj-ZNzyRdr5&@4}O%3f~I%xVa(VwxTqbNZk;F0zGzC!>J>@u
zP8UJ8%|@<S@+c`EcN_iQdXe<YYiXUUIPnbAqkW2EY-7wbOa^Dj7{85(QXzgDJ_(`k
z-FP<hd>jhc2^BxpiEq0DWIHDa8pr<-=FW`aq-$=1VNN!u5EmiPduheiubIW=c8Rdi
zmOaG!_$l(X?Gm52ohYo?b(h}L>*CgCq~Nln2gqh;U2M7f&0)Et2Ck4W#f^KjA-iJ*
zHLF&^ftCxz<f;u3>9%qZvF8M{GA5$Jp%S4}P&&?}lSpSnGT&i2gD3jC`TZ22q_!sM
z{~AsIoF54(;qIvRVi^_`@En1^)#zY7NR8jE1O2xja6yy;%b(=O@33aD2oGSOQioz2
zbV>WT>$rTJDjV9WK&6{cLr>ocvYbBUEaX;_PEQ>$E**_C?PW;WT8j5%Oqk2X1_*z+
z&f(hAQ)qhO8CMhfgVwlM2@2O=#~mXRp)c(@Y`iQ&PHp)J>msVLQ}-wB%yA~#E2UXt
zR~}~Gi2}E)*~BO7C6*eECc>S39zc5-L(C>{Iyx56sAx`xcPG#)=`_wcOPTptj3dfw
zh3Fdf8~+?Risu}8pKa$c?m5rD&W)H4^FAoR_tBfFmTN8i+ige^HW;wx>iY;wR^z|y
z`6Oz12D^DQ5(2jt&`}3d(C3mEKbLYPjUth#$IrS%%~p}JTw_+8!0`EpdF+ouIhskn
zMvYG@END(Ptri+G{kf00z$yRX@$@HXz3wk{$~Z<QYX9Xda~#=W-7=^?Z^)LGKcZa&
z#dM%dTd;=j;4Ith&MM6csjwg$A1wF<7Y*w`md_a^+h$<_J%)V}dD#BTnvTvq%k7qM
zLFM{(d^R8rgJafFv6cnMZIot*${ED&PeZLE_fTYR@b$oSF;1_n2utp!V$Z+<GJ9+i
z8T3BE{khyK*nPMM!oHT`jz|34djDuFv@1gqwK~pg)h>K;GZ)?cOj%-fF$VR&r%B`|
zX5Ko6qK7lkYoP&L_1cZx>h-Kb^bkkK&m@x@^hwZcW42quiS(CELj~UpoS)+sqM+M{
zvHI=wu#F|#vsjsbzjtC~UMy+3sX&vK9mKAiybq_1XHcdMp#mKVKiaPHo~1xkG%Cay
z<%96PXADc$lwoI@&OqON{%)|x7O*XqYd>T_5`At1ck~Ko)*7%-RRz`t+AMHTnKY9>
zptN@aDrCr!VePAE9+!ew&gqb}^QI)h<uqAm=!5=~G+E!bT&~MGj63wG9cn8zaO;Q+
zEcNt*j-v;8_FojKY#oF7gC>l1CXnr$2JvLaLWpv%h2G8kak<n!T>9CZr7wQLi9U<L
znrrTm!<<>{K5OQ;s1$!1D?#{xRpq($S%P1i2k?-I4dmA7p-aJSOb)YSz7|6;*362j
zcdsCQs^0`IGNQEV@C9Mc@&wW*`yCgi{Dg$Dp4jqQgSr;X#d50*7;(o4^gUhaMV`^P
zT|XZmD%oQ?x0l=FG?j>aI4JPE6wP_<9m%58yK%So2zL8a5$Ad41ln2iGtNdO*m1{*
zT|P01+`s%s`1IT}eDrA!I%fBAabql~Wz!5iCK`hS&4Kv4`xTb#YQP+Eo<k5P#y0DB
z!+q;&+_FjxHcQ+@Z31+wW)zG{a=;S{HjqHk7tnC`1fKdM&W_0{bG^0tBsF4#aL1ey
zJj$QnMV800>fSYn_eX`Gp!^-fi==UBxCL{SzK!3L_;Xa=jgB-HW#?6<;lDZ2&{4Vy
zQd%Op+fGYJm3y+#V@Cjq%3lGh6B&A09HMKldXw9x$rwJQNK>-(nX}#h7&;Sws=6o)
zn=*$?NivjBNk|#)S({1{5=Dsyl}bo+X*Q>lB$<+=3`tT+xM%H5MN%Q838g_a_(~-W
zzWoRE`{{D-*?X<`eI6rXrx*+u?APN#K_Y+E_!gHa*3E@ptSSp%;YochBpJ^(g={0U
z;J;2Kntt~xMjCA7MXLSbeuOBwBOi>B`EyC8!eoIu^RP90@p#uOn$wCs$0d#XjL)Ba
zLbvv6!F|vn**{bGj!RRC*7kHR@3c0({>29hceKDwu_)YEn1u#qirB>+<Yl}k(Vx;f
zjKwnzUp#-nPh00l3KT@~y}J~><1!f!ul$SYOPt8-h-iqCjv#+l6hp>>0DRMP0-rK3
zO(f$qu0JS38!XjHeQ7dpZ)65d;fS$6UI`5*)xq!Ab78+lDV{O-&3B|o6N$oJ?qc~|
zx>$+zS&s&zWExO{4R~m_1nJLQL=(5B;Dda1I=-O@J04wvb7`UYF|!4QThyshYd_~F
zB7=3F2`FJ7jbAs8CV#HBFz`||jB_3bvn$N#$NzN5i+ksBw{<V7wJo9TRcG*dxddbm
zX>j&ZdHfbFD=y^9R?>S!joutT6Ve@L(J+>E)4f;4FI#X6Cxoy*!;?4Y)}c&VXU-<o
zCez4?8%?k}&jYTnyaru~lemu$AHuSP2VisKG4FRhkxMcQCl&ib1?{@lxM4U6mT%ZV
ze)roDm%up?rf)=jJ1T@)lDpAi3J`&pJEr8?bJp_PQ26@{NIYV9`=#6Y4woh<aLY%9
zjX5kwZOlJE6+()9zwo6Qg?#O`N={368A+};ro;8o%-cAX7Jl(0R;Wz1;zxo%o1gvG
z{*E<zoAK#jBRcu}!lY;o`sY9@)|U0~R<F-t>4{3Lun3@SMI27Fea-D<OzXQcvNW~I
z9!%}W!wixO*B0Nx3@!&|J6NLX9s?qBv>ke0ZsJVd1ftgL{qX#=BnjO;&qf-YsQ<@o
zE>k2Ia#u;AW5qI>#h`${FV~>_`c&?C<`lSoIEPDGZHHH9nqYH^EuJ!3L5T{2@82<C
zRJ<Lu(yn4I>v?HATG2iIk)+~q8au0;$4p;aJbFxvI9|An&rU?+{?lQwY|BH8JsXRM
zL!~jYrH&he%H(6#W!}+lGR<#%z<cIbVo=RZe5=bm*pHj}d16!PWo-^>+{fWhwG}Yr
z*bQCD4H!`=#4Y9SZ2xMCWB1rnlj-vC+HD@vFCui^G6~qs@_xaiREhocL%i)jTMV~9
z$=~i(B_f5#xnqrKSks)wazBj!=jK6vT|)G0m!K)S8+bX*3Rv@dACc9Qr7=c&MC({F
zuRb^lS5#>ciGum0$hKQpm=i>ASj@o8Yh%fYX^Eifw2mG(KEgPHpHTkj7@}*y^Ql3X
zL9&&{jGq~3r>lq_w}&`AD8jLE8W>;WhRLtrz+ojLxEFVXo2+;XmzJF*A+6fPZh<};
z9%r7@7g{ttN10mD5<JcL;GUiibfcddYRt4oeo+ml_xT1sNiv}7N-q4v&{Mebkt%HY
z69hXS+-ASGJMiz?c|7dejbVGO>FD=wvFPhKvUGPOPTVkxK9=8%-9yQ6i*@@feFDhM
za|dBZ7~4f>9>+JcVqp1BKjLO#N<Y6^3r^=-ple?g-xrhwE@zfP<Btuv@N5oOXf%?x
zyuXR}^33TFsmBL>ohaN@LiibvP-TiS#(FHmeDUY_@k<P9D>cDd|H<^E_<PJ-AqT^!
z!tnj_c-}+iIeIVF!^yYXxGaAq`tnu;#A?l<*=8{iP~5{cIL!p<K4bjVJ)R7|EySR;
zCS03bKJGtwr95V#FK9gu;_qCrg!O(oD003<z&$^Sv(t9sSlL6UZ=phR{~h7aZX89D
ztH0np)~zIRdFZ!+eHXQ~fC~GvvU2{yD>aQJx_0h_3)(~yZ)&51_INaR(BM;V=+iy(
z=dpg(DkAyr3NA?sArohP!lR!=(7z@FKX(jrCTK_2SK84fiaPYYFpC_i+k>ULtLd1|
z!ywJ>5i#?h!z77d+;rNQiiPX(wFBOC%hD7iqDg$oT*j01dVtZ%KjFz6W4dnZGB~d$
z4(rXb&_+LtF%Kii^(X^a+&h!Mu_+d7Orv1ZiX9;REtE?(oJa-j7SLvG#GY{(`exH&
zqWh%)XLcIWK01fKGu}>2H{8V}o4I86hK;b}W++!0`3W0eBW|&NET~#$O`3arA=ULH
z=KUMT`(Ff7V!(P}G8t&Ls|;&~<caUP<9urUPFQ}e61VMR4w>WkP{=w`nFh%MmBuXS
ze5pu_E_mV-lX0}*P!N7wH;%?0HXw5Xlc1?R9miyqU`A;Q81Izi{)yLtMw~b(OV7to
z+ouvW*7dJjdI{#n-hn7i2{nX`=>4PugFX)8@D**EdG98-Ww?=FVs;0&Sz41G=heh!
z^J@r=*(~tylOt9NY0NR}%?%B+=cV^IZ&F^$tK^iE1Fq`W<+Xt5eH8=!mzVhRHg~e(
zmNE^hjzUw3!@@_0Q}~DXZh%42doJ`U6*Ncg1FzO_Ze5-ige*=bj$P3x7P5z|A2+~X
z`kR9_1xnQX;Z7J%k|M5I5yWcn4Vstf&|MjXtZnk9-5oOY?5eqNXhtATEmEbtHk);Z
zO$009Xc*ma6guXtLJ392&DL|HGvP6Iq{-6z;wM4ldLVy&U@R^z^uwxzyS%n_Blj)x
z8z!aP!;TVN^5uCYrhb@-+1>AXw@z(3zIq(p)n<mFzJItw8sgM!Rs-lux<DBw!sCls
zFfh=B>)C8t{=_wIdPG0W+4mBW{Y`#7@s0m!)&e5EZ-fhTmyj!|QCRdr4HGpwaB#LX
zyu7}WG`@)=HMK(+S#=(r8OO@&WC%%U+5-M14EgJP22OPCgq!=%qMfY2K%`@zb!w6%
zuGV>m<E9+}J<&7x?&Nk-=5dN_{}6>2s#WQkCH2^qco$0RL(yZbJguvE4CBJZi2aJA
z5aenH3za|e2c(=)WrsaU^QhoH|I7uIMkmzrIR{H!P4H;XX}CGZi#}a6mH3rR#pU;V
zxmQa<sdj-3aZ_!=Er9~iy<!RRzh4U7M>CfEfpIj)m4itqvY=PSkgV+;Nj;BCf<n|?
z3^}TVp|+_cNBjuRQ<b3qb^jCOuU!w?A*y^(nYM7{RZSuw)5w<{(MUg!g}r13h)f?#
z^T&-RB5O6vjBIY9Y6A0x9!p2wVj2xPpu|n^^Twpudfsc}b!;>_%%81Ggdx`9)G!#y
zi)=rI;<?e7bhZw6JIm8RB7-^WB}j<<WmLCafR6o@IQzvj#;RCGlkz=rY5!yByZ0D#
zURdL;zVqnktBECJwP|Sdpmpd6EZ1GLT%c6vgzje_!k!<~X)Vhw-nuA5|NK{v5?e3u
z@3g03#a{!#HR(3CTQsG&bsdOAeGC*f=i>|!W$GKT27}Zbxh=ET5{V7daeMDZT$Sd6
zMCvT)rzw&s|17@ig05in)o(a{f-$vU#N0-E@>%argwNT2234O$W2opzp@Z%R>|p2M
zS64moT>D4N_$ZAzdrJAAqD$z6=6AUIRyKr0%F|hg&cfsu1=!fqz`U3r(f@-ve_9Ul
z<t>grZ7d=leilS$^IB~F_XZa%xD8f`BDkubWn1rt!j|{ZxTDw!?cfKjnfDgmUOP~+
zNeTRHud7fjH=F-7=u8XK>X9&adqDnuPV8hI2L5@8<2#F(TO<PGo0{;gOFdCF)WYy+
z=0G1b0og@INMCw0c6-hw5$kr7IlVD3;H6AWK8dpPrzpNPlp_rXtoeaoXNmu~QDk;X
z44a|6hlkr|5vjsx;(J*c_I;QK+@W{qr?e64KWxDL;}8RHD$)HDu3_;O#-pGA5l3Z+
zlB6qEP@#DcUpeT*)rZQ&O;>}q^|8#sI04rkBSn_{cZMut{TJ5*R`g+qGS&;vBY$`&
z4C-8fgkvkPFtZ3W?^ux}wZ|wSWs0Bdtr)N65{iIGdGYZ^-rtF_|F=Hm;z#IkgN0Y1
zQc0UQ*5#A9IvregV;aqO4<`?A<gjO*<%xULNQjLy)re)R{P)lJClfC~?*TXb?ztFz
z{Qp4smnYEu88At@fX!i&xq9y%IAVV{2HV_(%mir??%;&ArZHfChPh()CZk{cQ#|IC
zjmKLHQ8~f@GN*5#b?Xiboe~2DZ-T41^~=|iruAc4hfj*!v|wJ~8Ta7eh^zSR<U$&3
zJq?#9pTxcT>#_X#W!Ppk3ap$|!KKHPE}69+qa|wiM7=OR$G0Cw%8o}X>q{_LcTt!b
z?FUzUEI?OcBj-Kd4t3e@{od9}Ua4V7*vz{2^0k}k_+2iHnYM`kcyJLleqRg=D;_dO
zGGj^j#X#O(dn_3mge|7+uxI%#u7zI+>Wl?{Se*4i8b<K1U)|#|q7gGs`w4dpGcUQ)
z5O?a_TmD^vJ@a>7<WBp(;BUsgLZ1th>COB^ym3DW{lgFN-MR&MWcf(i<D19#u#8_<
zN)u;y*w)6KT!Aw$eTmW`FL?EE7!EFQB43O===?Ey7!Z7#Xr5|EyFQj#J9P=Xc2DE1
zs;aT8PZuK(WTDw^Q#=?yimG-F2`3%&CM(vTKtVwq@fsr(1`o!fY<wV*bQw>s95kgt
ziJDaU##j=2V;U?H--)W>r6`tp7>(2Z!+{?YsgCn}l-_Af<A>*g%pYB{HuW~>-<?SY
z%$Y0K^e`F!E);Y+c8~xkOOy!HB-gFONNpD544<38)=ht)FxLkkPC#;)bvn@eF-)me
zq1m&o>Ec<#c+&MSX-R(sxqHsw#_Dtkw|oyGSSEP5CI~;>@F2c#-Uy=0kD`yx19%g}
zSWX`pFS6f+X6DW3maOoAJ~mf#Kch&kEEFKacmd9vCr9M{mh!gqWofBeFuyHNgF1BD
z6Q}L3(XT<4W);MsPnsI(ke0*<wL}=aBTI~pQ}N!zGayp%)cQe~DEz#tOspL`c(Lqe
z$nv;}XCr-}dxarwS=a?@=6LdBM;p@&g+2(lFF^y8BSBvEBSt$v!6}z(@m7@%eZ8dw
z6Q{)U!IKZ-wC58*zET$}GGv4Y&z|O2wr7Cky;0<zkpi(l5e){Td8CQwu;acE?6z;^
zMTSMlE?F}g+`51y=$uBgedm~OVJj}bX-MJ~_Ca0p6t1}0mKL0283HLQeC|00dUNfl
zq*(&GYHxtK{WI`LttuMjl)>EjbIJE*J@CxOkB<6Ph7~J2$|r{t>S^#2ZT@fsv=Tr$
zIgP(((uJGuR-%(vKK{1;gc;*0-=UDrXX!2xzGvS_6;pa_{(j9y55-}=<$eizO-jd}
z!Aj&;pTgSFW2sZ<1k7Hzjc8exKzdFVAG>A*?mg;8OV5bX{Uc`bqkhOS{%;^{Hu=KG
zNkwq;j55)DYck=&&B&tb;&g5BG$=a%M%dr*4~JG91N)wz9BFum7q=#nBF5YlPcX;H
zKN_G<<p}zg#0XT@OHt$2<@{llX>e$z2r;*d!uyNH1Lr3}&3CYw)!i`O>t_cpikM8l
z&MU>V`e<@dq>*>&zRbrSy1^;=DuPezXc{utkcQdTfJl8G|LM6By=N&wg|X)`r)nIP
zig}N2R+9V{e+7P$-g|ycwjNF#8N(IY>SM$2&%9M=I$m{NhSA&r-Z69|_SGizmsmKF
z3H%J(3Y_>BV_)*t#1aPTW$6CiIb4CI7rm<<3InT$;Io$~XSh3r+p}6g{mwD=^2fbU
zs&*Bt;`d>HyFCB*ku=DM1Y&gcci1cao-<Xc#fhr=<S0ASoSg8Aw@~YX=edC>aglK(
z3rB;xtvKQ5>QR?=SCW3*pB4m;$H^kGy#MU~VDNw|Zr#i}%*yfHu&pE=ENA=Oi0vf*
z@KH#kdr6JgYH}>3753~N&G=!KWd9G=ujtj~1E%xH`vS3FCr^up7zg;oNB&)`FBN}z
z3}hx~LwD#`3@PvDBgY6q_V@@YesvNA{7b+fxkKct)HM42lLhIzJeIs;Oz?RP?XbgA
z1zqki@4^iuXjmGE?x&mZ=(%7%Zg?advoeOpLtn}r+0S*+W)hM9nt?N0!!h1&1nl}7
zhw~RN#*PO!A<N<ls4}lgRl-y%`~pP8^d)yVND`+eNMP6dAYtY<WiC|xGdKB71~@s_
zVdJ)&sQckGrrjS;e|}v?QkE@&%vB%w(cP=*%tx~DmiaSElzZ`?_*pJK@w?#O_d)az
z^@FyC&wR(}OL+E1BTRPe!Ao*SvAym-ygTbnHk3%=#H7FMe0-ct@%zP>Iz;jh%Z}qd
zmHV(=T>_LgCGg`)mJ;4O3D*rj;H`A(g^gE|P&#Qgu75k5luw_CpGuf(_f$Q%X~i*|
znQ$IYxr&n7hNoQJ=WN_{H61T4zYD9Ms?+Vu>(NZE3Z-Aa1}$eflA11xF>l8b&$P{C
zRa`5Y&2%Q`-!d<wiXMq)9`$Q7GWb|KmBdvj(68E^xa#8(D7+j+bURK6BnR^Nl16{h
zTrEN-Xg$V?>>0wCfljcsDj`90r1`fxTVR_{Bni8|1IwRE66YWl6146p-bnBziN5#I
zW7#6I=A;rS?pX@)4e#Mq*ctNThAMqHeJZZ7GA9lD&T+0ekx=x<kiA1UVE7bYqT5-I
zJ16MjL7TH!>eYl1nnUQ?vKDJStSGn9fY<u0PQ#x}r^-_Mh+5KnZqj$wzgqBw*IaZL
zve&-?w<To)ol$0Vol6GZF37}f@im;&@FM6kDdy(gc#bn)$Dy6yPGRDaY@zy%lSD7m
zp6Ipe(1>3TxeL22s6p5TG+&@Zr$5Yvm(M$S$MR3;vT+c-PE6tKh8TaTB@0b^*_U13
z2)f{@4N3BkArfzsg)<ig5OsxP(4t{PM6xzpPuZA-9@3{^XVpZkT3L)%J+r9R#V_#v
zoC+z`9z`$WE6}{xi_Q0HP^5KTIsa-2_|Fd{O`2D@1<dm!Yc5Yq>`rj@8y3^<&d>Pj
zIgo~3nl#nilFn%}B<({J@OSk`K2u>Ttu3}A)lN5IU$q1coX6e^cNddCdDbK^qN&{A
zXfm(Yx{$hQNYJ@)kr?@xWo)*;!qlUM7<=tL+_I1*_sz?B=UN?7QV@-@7T)mvU=HX!
zS0V?mFQpNT%O7jq$zLu%3qQLvIWvVHxYThZxe^tD8_y50j`w9@P|SFm&E87~l$*J5
z+lh2QXEk%hZ^eS<RU|-t2P%#GDp+<kjKp>{!x-~9q$V>7-weEh7fu0qvX40tzZ9X?
z?P^vgV{GOXs`S<T5Kt3;g287f=zOojD`5(B%C3IyoB26@SJ7IM;1LZ=bZ62{Qvyhq
ztN|`EwIqV+L#UK2&d`L<xcEmugq{9RSugLI;MG1?P~EOY|GS_?V;BGCx-#<0r~Lek
zQ<k5EV%D{C39jK@6)eVrxL#Nv`<GkzA_K48ZinOM57FYmV(Rr*kBbv}<K-L|U<M@e
z{D&mDy>Bxv)?s<RFg4z=j7Rz1?&SW|SNvD!H>lb-l}2A3gAe0YksbGR=>FKpaLZYe
zYGzM_jI0=pyL%natyLmFMAzZ(Q_d(>szm3qGeCjh8b7+tjKtqnB6dAt@IC(-KG`o#
z=jXh_^*+h?Z2B*>Pkw+#M+tfIZ4oW8nFB$6D@f+?DzJ2!gj-%-#`71XiNdnI5SVQW
zC85Fa%sU>=#q~07lPMNVlx1GddR{*v0=BgFahhAkp+xy8Y#fZol4D=6L~1>E!`%%3
zPK)JJ98Gv#4?VJK5TVcJ3Ln%dO63)e`KOsR+~fEeblbygVC5nXWk)u_wb^3y(pMX*
z>Gcmw<A%BXtSiL$dMNK@s7QNumvJqMUpcG1UnqOF0g5zZgypJV;mtE=8oZ|xN@W<o
zYJ?irFMr3}Ba29q(*mM#{w?au%%cU(vp{x@1MW&2OAU7o2@fovMCzDdM&phzV<mf{
z!M_ryw_*Dv$DKr3=RKdK`il2cN`>;hTC|jxrgqD(@NN&Uat6Cxu=?{C#)1we|5PPm
zhy8t6$hy@E->1Q*TT|ib>R(_uYCX9*?hXuGUr!u0PcZLl6{_UUCtSA_`6?Sowsgx<
zs|hn<WwsQZC3hIS-Yuk$=h_M-=9*DMi<hv_BpTmjyx`rZs?qE_X4D+V)1t0hB<776
zDXz?c=%Y3GYwb&@nCmHQzoJVUjUPa0jk}F+VT<5x=TG!$e1*MvCg9g3h3(&^=udko
zsypul#GO&5mfuCm#W$<?Z4!>KX^bKLyy!ZYZkWd1y&+&3H#OQdU7TcW-43-UYx%-}
zc#s@6q5C`KXvItB0vbFowB8m*gdNMkuYWyBIiNr*u9pg8C;AbwSF!9Ka2D1bif5TF
zXW_q(W2k%vV@jp6&g9Z+jO>1fp2<tm`ByXv5j7@xVxM7>&Ro3JWlfE~bn#b8yixyC
z5ezKN=Mv0H_?N|NaI=*IjB;t?XMKynQ!mWuxE&jXBm1X=TikVlk!Bbds%wZJ(SR#r
zS=OzOCS$O!1}R}|EV=12v}sZ<_9iRS?vYaHu-z7ZFO|aC3LeBq?k8IPS;MJ5noM_y
z55w?DPik>T120_NN5rnb25r$tTzKetG<>^}G~~qao?bcdr?(7d21t>F88TG;t0kRY
zF^LS`&W0&VlQ5?GAl^~sxt%?(_-N`Tq8)OBTlpj$6&Sa($bS^oST9Z2tETaqdLq~<
zl1l=XjwhurEx4B4!^G><DD0o`pTIR@1U>OZ3+omf;=VgkOl7<tn}AsOsCA1};ylu0
z_8onL3yD+ZMQ(|gIvqH8f+Y6d=Ed*3leVs4jA62`gTJH6X4Ol?NuY;k^b<+g*97e0
zH}PdNw5XH)cjS7Q7hWTPJob{}_P;Da@3+x-YLI2YRUL^#Qwi$^SkT57cW~TPS6ci|
zffrqJ5>h42F|B+g75QGyA5lmnDZh?_;gkI+X-hz!8&99<wBy@nG2kb+o!|1ISFruj
zPVQFId9FVwP7r)Nn=^HH=l?yPLn>Y-;Q4>5)F$=-DezMQZSFIl@$V+fwr?UOaj7KK
z_6AOQtV@@lKgw5j3}Sc6Hb^SFh}V5q(t$0`kl8wdJRNW%R_TalIZcpMRE)cBj-gdz
z+ptK@m3yzThmbvmxa?^#bLO^ThP)Ql-SATAF*X<cKm90Qw?~T}<QdE1)<rOB(g*Pe
z<&2I0R7e$%@;lV8L(eG*`rB+huCxfoqN@tjPgIIbU=<>JU%)db6380XZ3&Q*$4tjI
zfoqU7xi~L}3l&$V8!njBe%F33$B5+%l9R#mq8gD-RpYw@#K@m!Y3}6wEx2dTV_yBv
zK`1zK4E8pifXH<t@SbKQYLzK7w|yO7_+SZ_tB?wXb&~LA^<DG{RK^5ONNhV96J1V}
z8hmnQcj1vVJzbnXz6wC~2Ad_VD#ABE_Te$hv*46^9Q02g;ydSA5;ORS<LvTmwl;5N
zS;z%6jIofun<hd2?Qh)8b8&ce8hfwA`yeMb1RQQAs?yoKoKH9=pP!Ejogewu6M;Zh
z{o^$io1nPn5t`+l2Ia~=#%28`$cmgnVpH<5z4Ru${yZ1+qE*Ps|L#MDc_7Iij>WX+
z#+>pV2{NtpCAX<u9+)?lNF7>A-#rTgQ|YJt>#@_wg4_M<Y<QfvtTZDMZO_5EGaX#U
zNaO1gM_lS%kFS`Y$!JCy&aEuQ!&QH=`(X-v&`>AS!%fJV5Q-un{Nd;kEn;=NP}pw%
zkCRN@%w~+mkT+Kwr)h{%&jfwK-O?hZdZ%z=a5QE}m{G@S=E|Hd1DX0NFnP~JnmpQ-
z<c}dJX_3S3%40F(z;>c@L6PPxSPIJhv++#LMEdjPXv$C8PhNiyg)7Qt<chm1{r-9d
z985XG?QJ~`VhOVN>)9Awneh%gY_36?yc>H?;y6Ro@pSJ{HvTm$K;e}i=;x=-sRwMr
z@;~)3+i4HRYb6P?9n@)YrVdTpxS#ai@gt%CJ_=L*<_R2EzlRwAaI`C#gd5h_u-){0
zaz?3zuVY=?ft@_eY4;+d9TgaROo?9ISqfrpO2njXD#n{O<IHN-+3GqcP=AvNnq3;S
z;_ohj1MVf#V%6N8Z|bCFj|lY$UxuyBo2&6WfEdLd<|CIX<BKjY9BbuFpB%dY6W&I^
zl)!iBqF}{sO5egMH7_RZO8=o#`9EHH>;S5olyI))F1+l?Y(Y-VbfTJ}1Pcb9V}RjH
z?hc87XEIYs!Y*q*zFLXIhH2ro&~Q9ddIev8no8bSRB%D8OZZo#21d>v4d;&UMh!M6
z?!UG`5VSy6u>743eH<7?ymhtcW;P>8{&5W^|MkMS;7AzzY&TYzFBO(}Y$9%F?}M=<
z6)x0NpiK*;h}*bxP?30ri~JCW?x{&IEVYQl{uP6Xk&L-2A&NUUNt1GwW()~9he<!W
z;epB(KGk?KE&a6-{$(7(KNdr{T;dn^{j(pno@az@D$9w%*po2!#uL2UXUaHqJHTQ3
zEMhz_iR`GGgl9);P=!Ud0+)5Q!kQ{C+^)16rLWDxihOJG$9)o+*`kD0^9Go&oq<vA
zBk7+<vc&FJpTN6xA>-)A37uw$(N9CU_~`is)EVWDhbH~TSnn|MqT&&3IGM!U$E7F|
zT*Vjm9s~D^NBqG*@A;neH|Q(lOHyC0KoAon`M;vbyw$o;@Sp}2ww?t)A0Og2RRR_-
zyoNJA6I`ge6_a`lVc183`#CH}(&vvMlmFhq;=K#W?9}mOa+DmKvoEAe46Lc^+;#Zv
zhd0<?iGtDj6T!sR2FFf5j*ZSL<q_iS4!HRY>i_)#1<G<%dhZKiirqxw6gr*${U8Z<
zyZyO}5w+ZxbIp+Os!_P9unrA`-}uMPi~}cl$!;Ut==EdT)N!RE3HZ>9civn;<1%Uf
zM_?|uUQ>uQ-@fs#4_xrx&M`!^QHaAglTp29Khb(}0uNr5g3uGwtlQUhaiPwM{As@-
zOvn%5=<i5!QcRK-28olwcO_i7FkAT8u^9bdZRdkRh3K~J6ejC6gYKc5LgksoI2cCA
zw(A3UvBrw)jH$$nwF;O%u$WVNBtsg@Tv1|SAAhqok1U$1jm>?5sMs@rmCtU0R`MO;
z&eyg?DT%!kE#=62B1-<KOOsE^3A~=|Jg!<%g^XUZ0;U_!rj2pCFwsMeOPId_|80xF
zRQH{z9UFnm882(3ygc1196@vxU2w&2b#kUvjjsH~_FI}aLDI{fR~xND;{$GT2?NS#
zp`VZY80WE}-vN#*i_&|=WAW?va&#PTPVZlzMvG4+U|+Q))z(#{b}JM(n-A}Ci?<T>
zXtE?jBb%_#Zx^iU?-W!DC)2%hx@56gFIU5|u_C8O*sQ!MO&5z*@g5`F_}l+dU|ZEm
ze!*@+{scFHm5vFv>c7N~CwP9~lLvG*RKu4c5%O^8BkM*i<~A;JgO=NdFle*~to~~i
z*k!-wyS5V&5c-Rs{a2i{3c?{y;THS+T6vG9yBUXTDedZy5SHxj#R^*|j-(%hNvgRZ
zuUv!`+6(Zi_BVW!7YSPpj=-Q_9rrA%5xf39M)BHE6d|Xe#6Xg)XXo(PfYJ1eE{|u!
z|6%<6Iw<&*1+{D^;az(k``3hGTC)_~)L%)L<QIV9y2E(gwVoK(eaG0p&%tf}D!y+V
zkT3tK)96e8;iQXtgm+#5+Qm}jV)${A*UmCkXC~8w4RL&E(qyhNtqG4gT_h4Kn{n*w
zcNmbH&KZo1Bn|beh^f{}3@z)kvDo~COIYbh<Tb^Jj+G5a_^?s-Monsb$P9H~#_(Np
z4B+q$E7G25Ku&c&#RKMT;5VWS+~g~{)5@c0II=wu9^!AxT|>v$4k%LQQ<gd3j*Bm6
z-q#Vgd9QENxeZZvD9rtYCC@sU|LTMAavS?MHe$pp+K5x{Y=ht3)99bgA$<Sw{oJGl
zj7buCg5)-?pgVq=(Z}m9<IBE(kQgUPN{#MeXxM*c*()Tdh|=Ct&5X0?Z&!?x&U4A<
zd0)VK_BL1;qfY<kD8Zx7Dk#tWh2|`q^;yw_%bU|8ytX<F>>3>ejxw+C&<f_wxO#x?
z;{>GZr8&*9T!n=R2gu|wMQT*~3I~kdpyn@I7$ac+uXDA6Hupj-A@1Z+RWhgBc1Uol
zumW4MM$rUK1+Iq^Aq~%73q#f%#w)*8V8bOl+ELghh++JptJ-p8i|-8bdX^DR`<6+x
z8gp^);AJ>kxPTu0HJ+-urJ`kA4&Sc05+d&|rIR|d0f(32bj5z&JL3-Os?-Y%6q}*(
zN`au)NQ`|Cf9JP+{9#itR})R8bA)xgI^B|Yl61+9rR!toK>BJmnpw=UZ(+>6XjO+d
zZ;6tM-w8ICZB#Mun-J|J#G#~g71-_m!<GKl$Ei*UxFe;P3zd>b+5H+^N|q)uoTx*h
zo_C_$e<t)=;ACFXG7BuZ5!@2CD@(iM05g&<kgB*U#vReYjKMpMFQh=1OjN|NIwjCB
z+L>0#)uL0RIEtu#;8OLv(2Ql(a2n$YWX0ih+s!zeIYC$KSxv=SSTBEYA&vQJN(a8I
zAjNT~p=0qSnBS{G!%QDTp+i63eJF~_BSunN*)ne2=XgB8(&fqSx4HR;ndkFT0rqrG
zCoVO5{9d=C_-0}jh8q*&^XM7Ie(>QgJdwoE)mJgte-Wt?i^K5eQq;WZ2K$-SY5Uc;
zoK9jknx6m2KiJa2zv@^50~s@k`A=7<IdU6R$YeUW^B5M#PQ@1I0p2TN8@gp9e)?X|
z|6CLWVuspun&CIJwT#2Gn0_3WD}kl^qj{sr2OxfD9L$=Kh9bW^g{_TVP}f!{h*$iB
zO}iEndy`RUDR%*f6n4Uibx-+@Ki7~{UBHgn51}I39c`}U!>1#Sg3``?eC%>{2>957
z=8b*)V1zimvL}^v{E6fqe2B!W8yaAF9Aj?pSOZlhNyLKB!b4RRcuP47OG=)g?52^7
z<M080btu!~ENfbJTY;8-PeY3hIlRBF5On*-P%*uY#Gq+B`E448mo4HjZlw`Td8tCb
zb-J+J$x&hc4942JZvg7bRyg6DDESii0JUG}v-9UAemHLvs1{{m#f=Pssb)BYwKK2b
z3W{%kvW%9T6)8LY5G{Xv2Fd8N;1ys%>MIxHftg}d!uAW7ep7)?iXK7T1UI<V=Yr76
zD@l<1y@|Q78Am1M3`w~741zQzX!(3OVyAx()@+p{gMQ6i^~S$!|HoVe)xWT%(uZ(!
z;~0}jK=#Q=)4kX9uy^_;xH}_->owa#DyB>pTrck8_78Ay{hU0C9BwYNlG@3M1(d+c
ze-lB0WiuQOXoB^VX>^&g1KxkX2Q6<2acy-G9<Low3{)3F>hs@F_wES)dA2{?zS9AB
z8LLHQ)mZ5IdrUCWaTK-lGlJjy#OP1wGBmlqloaf_z-d2ad%jy$*mcnvhVot^Cm01Y
z{}l1wq0DbsFamO4&!O|n<>=a*vrw}-5bPA}g;mXB)Nhj#9WSaux9pz}ezl>T^2}s#
z^K}<YS6M?GZrSp;_g%sb>x*%6Pda(<z6*<2)M5?GQe2*Bjb5Lc(O)+T?yVibWv!Rs
zubVva3{}UNmG<y?_b(W1Ixh$!nv@?d6XqFQfCcXV;WO1iEQvISTD2Nn4uyOQ2Q(w}
z3alC4%>R>(A$Fth;lK!goEtAmr+%cIl3J_a%)W={e`Xy`3%ZMHFp19idIXQDjs)-3
zO;~ebG70@~$0pMFGd`%lj0e)w`JcY*UFls@UUBkGx#aNza&;(*uMJj&Lyyczh=>}R
zct=6vwXd-2*ddI4lr7MFz|KEAf1;HB1gih65N%S}-<a7FFzwc*xrLWW;A|ayu<Qc<
zagin;<Lv3X_+@0nCdRfu>;^x$9!Pxpmv1^Xnsk3RC2RWA(PWb@d9CpnPdZE^g>F@R
z{t0!eyVr=m?B9fGqYgsQ2TdBT)6b2wd%@Rvo03CMT0vmHh#b9s43{OF(>J3y9I+}E
z9DI>+EKl>L4LU@m;Cp#VfIj&&&zoG5xr#eBPJ&~Pvlu7hHgC|?jXP6JN$#S1;6I}V
zmwrtq>k?03#nddU$@Rs~87idy!Eau*DTg<$tVd83q2}%;<WG?+z8FvOt9J=1IhfIN
zwFP|hvRv{zP@D$aj>7SizQL6JfRX9qAi~}?dADR?M3@Y<Sv3Y`1skB=ybkoueTkR5
zbvUiM2Cj5YCC&}~1|o7Ae8I~TT>7$D9B?{{mvl!F_53k-rA>x*H_Zf*&4xnLYL=bV
z-a$f}RBUEFA5DWVv5aJG1YU`cK&uH-DDl<@okP7bMC=ySt6s%x6`mx^teFe?)`20B
zN_<eY1#dre0Pb&A#U+wWcy-@)-1cKL$&3CdjGh<HUtW;U^-UX3uannkJh~9~e;7xv
ze0hWoC*NYtpBPS@dxvYrDbj!DbvWh1OUx|I63k?0qZwh|G{q*I$ou@j_tFb!>!*>#
zuizp-d(TMPc&rjcd&V$V{7*i<EsFQb_254H+~h>&t5HafA`RpaZ_d^6NvjA+Grb4n
z7P-?et0vJ;Hq2ud{0ui(q!O85J5aWw5w)Jx@}Z`YoC3>K$LyfopNnerqEwwQ;6ez@
zdvy|}2eM#C-xQ3oZX%AJ*Ek6|35*^6lpm+-%H~UM#QUl#Ee#)FJJ)NNIYvPkUg}L2
zh)kt+tH+SW2|)YIu3);}R5CjzifGr=LP~xHw>_kVGg1*l<#WNf@Y#E?+3CPdId}<-
zXGGv&<9DG`!*0BFL6ST?tx44{$6?pDCYuvqm<MnuhO0Vy6Yolopsx8W>-KmY{UNDD
zME`Ph!4(S<uU03xJ<A4t?I&?DcjpjgH&G1tKgXLtkp|6_`9yy9Ufv<X1Ou}j$-2~o
z5L4MLH2p0>mmJ&<3Cx#eQ~Ut}e7v~4^mO>b*zN^Aqo{8Dd(^Mr3I|p1;ZZ3m*paXZ
zzrGDbna%d({T?g0?_Y`6>~&%A_Ab7-T2avbW+a_9b_NP>CZfg+17cvM$(Ro6G{7>2
zXolNSE~Ogtii)^@2?|)_Y=TSo|AFNZ;dn62iE0;mL2$<mLZ8-gQ@_Lt62X8JxtZ|0
z*D)5ZOBA)qwT9wpdR%*43Ag^>ebjRA<#%a!fI_GuirpB4^&K10&^MPchWt@W{1aS9
zGxD_l0biKRz!GcEkl;BNiS~F+lKS61!nf#gbDbn;{gh+4IpGUmQLiRwe}7QW8!Sd*
zUYhWEm20qq3qe&$1Nc<hiLqZwaES+F*1eS^nUf#mrZ6#@E9D74zDJVcMuHI}9cK=)
z-PE2CoS>vf9r|VHVZBlOAs9=IIxq5}W*czXpRe2=cDG$}@ih+kH{jsR5QJ~6C)J|`
zF={6vDWek7oMMS)-7Ii5$;Ew=r*LD{R&0=3$IWl?CZW!wsfU^lc^za1ZN>i}<%2ja
z*nAA-YHPsw?I_MI;t>}T7lS?f4T$*oZv0)-iV|#Ambz1lQqTMNtAcs6M^1%<wqIb&
zTs0`{ci@YSv*GKm&A4!p0H-hhjVoB!?V=>xPu|>zg`d6AH~cLBdzA*2P&>n%zjEZR
zemp+iYeuc^8Iu!xwqj2dArrQ|!G1FxjGgRHCRE)*-3fABMVvL?QhA@huJ9Sc=O3m~
z8y;id(`D4WR*HK4v7<dR!tw4p8|u69Mfs{7i}}*vX~I6Nhs19eAX=pxCm#5N4X$U%
zu8{F$SNa*)aI7B_Qnk64$AwVQ_>6z$nLxb9PQY|;S5o0VjehoVB--;r@NOIXThx1r
zH$G>A_U4OxVhD37?3s%9^Nk==+6yX&+3c)9gRW9cXYa>!&gI}ZknRfS9n{xhl&dzW
z_f&%#sSSAUbU9kwYz9Z^KJc7kgroK5(_`~qqo%YcXeD;wT53l1^JK}kcIHiaRl--T
z$-@1oZlledTILC2b4{gE5N*GKieE*ryXY{=3PgxZ)k)lvGn!U6JwUj#fvW#bB;3#S
zQ1<yEx}DRck^d3|8EZ<o8u<|<r)(R#dyODrtJG<}wh6BE=t0LsDNgUjQ?yz=0kuN)
z`4O8sVY+21*seaw_k0-1oVU+V+iW9ajU=J{Gd&XP+>I4FkFfIMAHGgYh}}{R{6Nbr
z>fIVjZ?_6)dT1jQe3&jAzTb&27=Qlc>{~E2)0s6>65#V&Q7Apy%NzT$ipTdY$n95S
zUH(2!EbJ*?5WNisBnQy)?_~^FW=<=a6F_C#Jy_}Q4tBGegeIFg5_+$`TsZ3p+a0jG
zrlln<PTb1<^-Te#UvGuYcYGkFlkq~^Ukh6t{-UnT61vQ96m8j7gwo38g4eq@lAk|^
zFyi$Z{Pxs|>gI)za%poSV;luzk~FAw;~5eVs6mgtu!64kG~Q#tmb`j97K=m&c_GL0
zD_;gN*u;hQSZ}70mCfjLMUpz@ALobi4v_S9Q?b`ahlGBzL$`5qBv;#<HYeYL{)ayW
zxmi1LS%DXf7}|vELx1zj{R^;R;YoJqZAY&ohP*=Da=NneDLR84ZFfrK<$s5hH5O|4
zDC-Xn4<_*KYba*wN|C$j`Fw}99g#X?OeH_fqN5ALp?||7vgmmp#H;`27JQfn%6?5C
zb#*h=H&k&wdmW)vCkk6XIMcQRx}?!>4yX5MF7!mqLkT-0__{?MO70Zkfzg%NWoN*t
zZ+4+6zf`E<2MO|`*Pm|cD@ExiBj{4+@kGvzWfGWw*YBYL1nQ5W+rEsTA1&KZ(ccrD
z>#jj@HskGABthMRd!V~cm8yRF$-A8&fl{mGp`^->gd5HzAwqwCXmd8I$1WlkU&het
z9v6}qSIqyEo{svlBKWz)f`rUeBG<m|K?}w`Khd^=;36xcuw@LiHz#r;qr$C^93O#V
z$M56b=I!L#8B5yuvW2gz)<c!yWVn7@3q!h%>2DWv%1vzI=fy`tt;_{<V$7LEr#zr;
zn<tF-8-$IA(}}NTJ{C0|C+|mp#2q8t@IRL{Qta*vFE*ONqU#DYJy(&IvG?dfs|48p
z(1Roz#1QqWPpH21FFqt%w7sH}>vt{X`}WyE>4Ej=H~TNv-~7kN9M8aYX@h7r&IP~K
zIMG{@C-8=>DCRjjk$>u5^ae8^e6%UU{_ap=pMd}ri&Gfe$B-C)kfLYVU0wh2cv5>N
z1;4H}p_&T2NNDYDfotv&vh#im-g5p18LD+Kaph5B$@sm6wUf!Hu<O_pW=PIz+R?4H
z^)R{B0k4D`(^;UxSJo^igEJ>_BMsEal4b4e%sq;H9$U+0%h*zb1AaKpXaQZ5>4vjL
ztKvC%2{cq^-a)%2+&jh`KN&40y4j1VtvqvznyzMkn0i5G@iw$mFNEM8ak^9WG`{~U
zLgkL^LG2Ou(RP~vpB9CYXzgT@zrm2^=-6ORkP%t(J|4tOT#1nJy&anV@w=-WNpsr-
z`X^!&$R1WD$@WFKOl%6|2bqG%+xOP1TGvvyJ-4_^rTuv5&QkX5-NOn^T|Ch~78S3j
zVfmhqa7SF6zDjt8>%TH)L&XC$*d2=vA=TL3HiEv&?#Gp{Vxdk^5x;)Di!TaR5V3X3
zIPYI^K&Q_mPbx*}shbk)`7ow>`h+&id_?o244C;M0^Qyez~cky@cl85En#b+XyRId
zXU!_`JZwc0ZnNB{ggxT*o3P7e9+7mqK>pTFq6`&@7r#v+F*alHlUh7hTyPQmI48l4
z)To2L3m3VT&xbK7W&#ns=!RXAN4d1YR^(PK5%Rr^Q)vAfJF_fl?uleveEI~2ojVDl
z9?V^$91f{%7hvV$K%!jXh`-|fi0SZo?pv@6U43CA1Wy;I?Zu<W2Tx0Kdd&lT>^&Jv
zSspS~$rlZWIcl7Zyu%Ms@;XtRY8yr1_=QavawrzYHeZ4$;bp#zticzZnsm);Wn%f_
zFJ8~f=lxd;`D2oHpz<#s%UxyZl)Vq3k1HasbK>yCt1$Gvk`D7E9q?1$4;<Cij1mcI
zbi%Aa4C;KvU%c9c{en!Q|ELcOOt<39aVZ$$V2$r`<)}>iQgnMMCX95rhX*v;uzlHK
zvODDvI(+ECnm`8#V6)|w@}D64lnl9kxdi)v>=qXKZsOkO?BR2k$dZ)@E@6sSA#=1$
z0g+#)Z60`xCJmOyIR&XH)K!tdHkAMp_*WIBQ#$yUxv^OEBZu|ewiETKnXpBr0-N&}
zlk?Uexb%ocxH8e0T5L<<pBxQGZ}&UcXe1z4|Ak_JVhm%VE#wrGr$AQlO@86RNc_5V
z47P0_hco}wV$R!4;@aVX2jyE)O=b#XEQw=E`5~SLzr_;mJ<P|^#sAxs!5zHY3+B2}
zWS{peY9mT9X~+{NOSRzFS(3zKdp|DkzQo&)zJxXnF0{I{fRFp$je)bJsLJH!<j{mz
zusc%3=lM&(l6j1q{^9@`c=r%@d{INiMXKoV@eRM+qLJITs2VcG=Wz<JS7559qcBvl
z(PndKE$?j~O<MLp=cPSXa^|uZ$s(z8J{A<wsAGUR-twVflm*gnGl|zb4?ce6TD<H&
z1+LJEv|%C7c4D)Lu3o6%)0^$Y^}I2r2D;IQ6Krt(=8<H_d0+a>))eFp3}auLE6X>q
zJemCg{z|R_EeUet^CW8^DxbN{n0soA`2w_fb_YL&In!p*XWR$9H5fP5fWm=m80Hs%
zFY><PZLLJuQT#&S`Myz5zC(`6mxf{Ze|B&ta0&KQ2XNJ^Tq!%HW8-e-LXT>||F+7|
z<Wj(KV{PedO&c0>D+!fqg2?rnzufM*k!V$~OBbRVy>|FC7bANCt*6V<V<s%)eE&A=
zjS(e9672mZnaI4ZBj~zq^3<kMp1!{Qo_}k@c0!RIpsjQg*BR;38`tEh+fz#{sG7<z
z37JHLVk%I)#|*bE>}CutDQeZ?PwL*?!N{$Z;JYUZ7Y1eE4MRs<oTNZCtVIO1!_(-S
zCu=cW_7%)NAwp;DIspdr)<c`ibLiUgQc$(of>!5b@L~zaVSL>!IPzsY)zDYxpWV_Y
zSGi&gmD$WKeR7@*R<-gUTwm~A$_0Y;%bgJ)tCAb~jM3A63R>rF!U-!)q3+*a%!yb-
z>vd8ft6~={6j!E_(bdFL>^r}$?iIY*u0kd{ParqeABDTWj3{UnV~F`&Hs3r52|n$p
z`r8eX7_TLs_6WQ_`QmbyatOY67WQ4ZgTJ1)fUZD_CZ3o`vSlhT*XK4!8cm>)(gQ+W
z$^F9b|M6H<>p~^YupFO_6a6drg*H#Gvc7o(zI8hZcTXwe{@g^Yn=*se9BahT>>(R5
zi(YKp>50}3qiN%)ew_V73->)eh2GvWv_4RW-k-k~W2U%Z-gPs+nDy9-11Hh=q*y+X
zaZ=VPiji&6P54J`HjXn-#Dbk)`OalWV07nSESUQbW<SrscdYl}e%Ao6cTS+ikM7{|
zF)jQcrRb{i3=ISq7-y^z^W)^nhK_gKS&cDN%BK|Pf0yFD&M4uc3CvML3)vag1hxk$
z(9@5lXy||c3AMKT<}=%_;=H2qbW=qp95|H4omCz|MKY@e%}G%L<D-nBV3>tte<)BN
zZ&7+&>>^5dMhp73F&>TSMv^%YELi;MHYzx0;c!_8h>v}buMU6UWon{e|AGTx@%bF%
zIGlk>vnh0Vc@pRB83yT_XYv&rM+<7oF9@<uT!w$85yZ5L%|LIhz~D{)&?<%UmHU2U
zOPvjTNPZ5PM}L=ZUhm6k|1iR^oNF-f))+q*HDmATzx->~2Y=G!Li9(ca_*i5khXUo
zZ|uls`t@O0|1gZ}@?4GLw#s1m=Pl%G9VeAxYNTBI82&K~1<#IB;`WpEmey>-q7VbD
zoFqkCC)T6V(-Hja;vB?b)-9r8#OK~R(st%H%vl&kY^~YnT(J@yqS^hi`X?%v&x7W>
z>9FEF>nM1JaHq8o5zW9SP!e$shtsa$)*a0GJaHP1Jt#%D9yj50OuV7+pc>&XXK*6>
z?$}6wo6lxv>R6ZHPtx8S(x?}vv^ci~roL06F}Kg~M!#Y(<9r)R%xQ-|`R*jtz(}|f
zzu}La#h958%H2MH7bO~hfyRMQSmWtPC#^gQTC;i}`@nUq>(>K6p(eLT!;>f`sgu|5
zmvMSGMex`*)_+Sm3juaZP*wFheEpV;A8He@@a||LIC_NGJPbwM&yvt%zLp+5Btt~j
zwUtdy_rXlF<D3I$$ql|sCAA`Tj4^NqwR%1IqAe!`b*>+{*{9}E`(NF{AM>xE-;awp
zzu_(RUOfw0K4n~dL<AWXcmW=^EueL3*{HXm5$tSdQP;bjxVM{i2?rgBu6Q$>**c=g
zAy?j^xtQ&i!ZF+9JO+8ZL-Wm#;E;L-6pYWn1I~@uTWvvNA8M2H!)LJ~Se_p5V)@U<
zb209s1ip@R=ijC0;IyK@=(9V99JMIpg1_gYrsWmvxOxY_4yusEnR{S#t1X?{B0@v=
ze1?$2W2l$oOulDEC_JVE{OFh%QdhehRg6{P-O?ZY{nqQ?>i!?dCJ#cqU^ND<FXsBE
z4)Yn$p5TJkCRjdyA}!f;oOtMGkwT#}t&A&x@T$|Wc;h<i+ZIBC-+R;6###(f>%);<
zQlw45^0bqV;L#?QyOY>UW~WW1Ieqhp$Su3_?hGJBSB-=+w^tA)xy@u*hz_j{F(uYn
zZdCs)b7%V>BI-N4dBK#GxFGR246*LE)`8Rb^Lz-IEp?7FU8ID%zQe)*?d8O)TMHYO
zRiRmK3Fqnj3-<3!L29uD)`>7bwn!6BkJmxJbxZj*r!DZ~3o-KWjvdy|@*&D9OUQ<Y
zYjBff8HRNxG1qrJ7U@?CWc`Lf`|ENXF{%dCT{mKkts-YX_XFnNl|dz)VzikQ#%&du
zL!bG~BF<)!ATh*&>;T~MJ>3|-;w|ooD!?wrkk7p2PV%?gkwn>LsI^&&p06+>`-e<0
zD)0<$s<;ioJM?MQr*PiP=_CFJJ}@+$eYW>5@Nd#zVWOxf_!p|u_4`_wKTU%U)vx6f
zTS~}>|31UBB6(taY#1H)1rrl)H4*tOTPD5o5Vsz9{7>&I`XB5NhKRN^#>P8X*K0_J
z)=853q&yruTAqm9J7?|fo60|Nn?fc!w!&RA#ufc-h6=y8qHeAsr|>^TXC98#*M(sr
zL&gx2DWQ@imGGXmlS(Q{ew0LMAW0=@mJm|Lh*UC^L?ji3_pF_SOi7YTB}po&gyzz>
zzrS5qmw27C_gc?$-xYfJ?KtZ(x#tV(FZ<)U`XpY`@-pvaZb#mIEk+Id9--Vu#$iuQ
zMcoC?=y2{jR4ntxFwLc~(|<K?lRm{I-IS%xWBPfKU<*E7(1+WKpMzc7SQvjto0=Sw
zqGX{0J*1&TLZ+%v$1o4<7$Z*t9|U0NwM}?UBpB_t>(QL{C;Z6Gcd>9?KbUkKf+tJc
z`Ac7iIpt63pqTavZTf^5)2dE}3YXCn69&N~F<zkHJ&sPF--D;6YGLT-M-*+CNrsh6
zuyUCTk&<raPn-<J5J*EAyXj=PiyLZu98HC1Mw8&>RTyJm%Q?5Y6Gy=T;>t^L*DsF4
z=^1-SzxX-SV>>dfTW`pOhtZ_vt|gsa`xXtmSRR31Kr!E`^w7zFkiPLSrtW50<e7D7
zwmJYER!pQr8g;nld?%N5K#acMkO+nUiP8y5C#f|S^Vbq}Y4#y?+Uxv>e|$I}97Lv&
z9<ilZ;2=&Wu3AXsM$e&vDw^cvqg^;<-B`*8jpVI92V&*)H@L*JA2W6Sl@B`p;^_Ts
ze8M;{S|zVBu5TeWA6bZT3wXZMIE4Rvhv1gU8Ss8hCcfI?MuuE|a#nFic%6<#%*CC@
zrJUyR($sv&)b!y4#sBh?#+pNUng$&oFa>q@q~q6L8mLs+iajy0TnzL1WXsiaQffy8
zA}lLr-Y!Z-{Dz@-iWzFY+RtbAq=HiPFVsKvkKf=lg$nC$!q}!ksEMrR?cxhDUu_!g
z+<#QCct#b5Tz$(=vUTJS<vEZ=$&X=CW*p0l&EdKaN<oNHfUr~Qdbw%afM9u0D70@H
z#GzZwAX;%8T|YaJ;BagDvBQko?MVeOyBd5q+X&7*h$50##*&DQ%Q2wSo(#V(N5zqh
zJ2TvbeFLp*?kG#!Ol$G)_&OY+5DS0x<niIH3;ab@1B^R04knGShR-ft`03mc*pXvM
z^zUS`c{bY%YkWo5^}0CZwI98`2eEFODz{!$h8)b>N2fd;O<u_Vzn^bG9<eS?!O}6j
zPi8T{{Jt&wEcIY*G~>3W)`6F4B6B^l%;w*vL`78zvyVSP(NkCO$8a*fxqlbtrccAS
z_e_uol=xAKiugItkJwbc#l0WjVbtJq*zHsVVe_TPsIDgvy($lbFZqLgcsy^uitVr3
z0^rZvG>E!pF1X}lO)p&-Pkx^oLEb&6;_VZRpeeBlQE?XC9R3W~+xfuv<E@~wY!;PG
z`p5rRJCzpeETx{GB<az!;`G_kBWS;L9F4lDOPlt*!13)Idmc|F|0Ff2^fX)Ixce2$
ziXXtezaQdhhJvlW9z=}J{Q}$QXo$LD!vD00!KUfS_}Z@@UX4>EPaC7iq9w;rbZ8-j
zZ&(iN#Y;%N<qQ7JQZd>%-jn>jv6`&W)j_kwVm@m|H#)zFL&wt-AocSE(k(s!ZbgZB
zV&Qk(^x_6cd9SByQXZm-$6NGO8{`+~j3SP2Mq}E-40y)4P8xQr`G;q=W6bDo%y~Ht
zb4NerpQcDrzb6HtFjg6&BSPpF{g==duRwGcXIQnjEadioptyT;7@7Bhb+lQIy5*W2
zb+TPYbtm6P-A(a=@7Z_wKW_J+=tVTHkQI<+ISCl#@tnEKx51LKDQIJI2E}}G$v4(b
zuQ_%CI1er6^DDspjP)O)+amZ=BuCdA7)ixjjtLe0X3`ZiEa}bksWfx?HhO2IEi5*W
zBva$qbBI=BGEsmW17*zgV4Q>|d9==&f+BRJpl5Om{Cv6$#(kYZHs6yXL04mllGrbh
zy&}rh*UM2pbPtM;j0A1*0)CsE23h~`1eud!MR#<1pl;_nmb*Ge3VL3^;R&tqY^OSg
zcicneWL<cBgE<(c_VHtWR)Bf@V)!z#2k4IFP+-0gPKZ>%(eNdd^Sj13FBau4Pt?H!
z<JHiI%ivU<dpU7U7pz=g1kDp8$cBA$u+(K3_=mNSJ>Ci>cFW=gu@gk<$WfA$myd>#
z*{I(ziaN7iQL&5!{dT(v2dYZ3-R&9vcJl(K!tHqXgBmrDpH1D5#^KMe_i*~)S<w7Y
z3`tUc7_ce{L)6a!#Oje>*K&!^f6-t(&yz%2PNoM>J45s>6Jn;eA9VJ8!6%lP+<E(j
zG%({AZzK?6_Jtq}%T*^T17c(a>m&qKsd5fVfBD+&pHNrzonY;QJgo1%2kK8ZLwzZd
z=@*pfp@ZXT^~cXB+oq4w0|sPh-ymL%Jq&eT(~07!PawOFb&U2MA=~I&F2vy}d}S_|
zk@KYKcd$fd^%2CNh&gpJf*-3<4fZ3pVTtZUnwFDBiXT70;*G~ScU5hgT{@53eMS||
zpA#XsZcL>rvs+*wD3?7W2|V`9!jDf!P|aV;bmQ2k@bIDuUR&jg@|mIhrXv<~SNa-!
z@Zu*XOglq{*uZ3feNFdi0A0t<Qsr*i)bZ;nkozwIOm1r8@9C1{TS6RnT5&bHYWx*=
zJW-*EtlQEzKaktseFa*QL$ISL0$rtUqNm+tF6G$?yt?xxI!#fb5|!m}c_m{+Y?h?f
zCnd=ftMTNr$RL^*)o`_2*P;JF3W;f~WX%6|oRF17qHIRdCF1@>ecD3Wt2Tme3gU=W
z?rqR3Iu4#oKUl4{k|Mtbtib;GW;ijY4yC+|QN&$~DvB3F#kd>7a``tPu^@sgn!|F4
z3yyG;<YLLF*DmNkm`4KYU*Uw{Ffe|YhmATl*i$D*^YxYJ57}4@|H6ErUrcFN-yP2Q
zc3HWG=?nh(^YL^|;v8CAO0nSnTg*1nC48tGcWre$+(@~FH*NMnldKV>r7BSSUtU<H
zt3hRB#h8m=FU)DbiHZt5^kfF2RkAJK>9~eL%Mgu~qw)2NCYTYq0f+V|;pcZEsLl3e
z*SodhbDKL<zKbOOlQl@~7nT#-TMPFZ)?(k<27bEtQC{M%88s8cgGZn;4XO6wKCXL=
z+4mG^bkS=Zq29+>v~TgG%z4=T@+9NwO3;tXqsSUlc^n<*K+M{Pu`nSS_l?tm=7L-<
z&oGtbL^}|b$qV5}hZrspGlAVbr{VNdM?AkQ5>uBw;Fm8EC)|;*+=+f$dVg^mPCD+6
zx?&fFa|WU~$KeElhRiCkT%tg*MVDlf0JPZ5SYZ9<v2N;EdMI3vw)tA1ugq57%;yrP
zT{6tAS70o^L(2r=+ZW*bavrx#nMLfoF5+OW6TJO8gO0y+9gPb;al<xKQs^FtGiRI7
z{@xgv|0IhPWb7jo@D`2_T|f$}tx3Am4_L-HC~F?E_n8{P7;!lgEE!J{n(U#eE&)BS
z_zAmiF2TMIJ@V~D4LIMqjCO_}(O1!h=)_GRCz&+GUr~exs6@a!kvq6|^;qz^sZMm8
zjOgde-C)c*AtD%LwK{Azx%=D;W;NfzDN7_lL;MeKud<jgezg=V^iN<yp)?)dI|w%}
zcA?487hDMs(77#^h)_+d(LD32{E#UBe#4$*y7E>t9$FIGEk^S;{zTjOBXD}20=+TX
z2Bv*%!`&}e5V3tXVVf_bb}EcO5l2b>@!4T4j@wB@R6arXerdFKEax}aFQSgqJ_|1E
z<)Yk;-KfU0R4*T7q3d~fn6aY{`>&`Gt}_W1Wa)tSdM&DN+QRNIW9W}RbI@YMC0@S#
z5hkxzLzQ9!=t)$BC5DH<xLTT*HETrYMdwjA<O%n5WG`MTpAL?}b@)B@Dz6it#BVQ?
z!&B2fB1huT?8AFZ+n&MqUfzQP$4l{Ms3iSY`3;{JJCIS24Cs`GM@%Z)jH(hXn6@Vs
ze<||3@vkg6DbmAPpQ*x~ZG9L#ItetbC*bl(Sz3`*UH(?{1d9HuzySA&xGJ3Ou&bx@
zFArS<g@iIFS1$zXeU%`+Y!)FdfR$y6^nt1f)zEp!tymI<vHKg)sJ0%0KIP#2ZO)8!
zQ;uumP3XAI-}$hu`8ab=FVF9G1FpE8`+1qM%ja%K-};qgzxNO5o0P}PoK1qajCrxY
z!h|-zT}ZdOJ7R+PPq<tx34RIbXmax^#9UTFi%B++viu~zdLvJ(ypO=aEn6_SXEl8L
zF`j-({?7YE7IM{$#UEl^41c=6z;(anWU90w$#d5vQD;JVF{wGUBA|({khY~&{8*~%
z<xswDxh^z~v7@zq9%O3XA}q5DKtIVOj5}&YV{$IS`g(gTAL!&&90&Q7v}^FlX&UPu
zuONlY2kt41Cz}7HxqBZ<kjf<D{l@dyWjls6ve91f+-aoX_<o%9a1WH!<wLKX7_FN&
zpFApDiJvW2fxYbu^gp?&JYb0`J+eNKtm5{<nS6D6dvX$nS?A;Y4KqOS@CEGY9Yt&J
zD1Z_3%kCM+9Etxe!Ai!EQ!LufKYxFPagfsZ0mf(8%Cd`@ms6oxhj}lmyl{E1B;)Ik
zB*8`r@FSY-hD-9$QI3O{Ds{SEWG!g!n1$S@7_8Ph42!fH@#)y<aD0>~eKGeqM&69z
z+hp$H-bQ))*j<cTZ%76!@d1=AktSBpoJsSKT->aYjLaBK=FWKlvb)^zTtzD?2E0HC
zxfHJZSs3b`9uTxBG0wHfNJ<Tluvu6ddd}4m>`D5`Ic|?7H4BVU_Au*x_HIVwo*_=!
zlwgA5Jiaab7TGDSf)_I?afe$x7RZTWu}&?9p85cTUdC|a**wTR|B)-WdYdn}^OZ{v
z)k4F<bQD-D1yYj*AzAy%i(M1>DQ@Rb%I+9{+)j(`nJ+>`;*83@)mOoF#z;*J9>PC`
z(R{U=Ip}&Sa&s2H$8X0z^NUZPgs3ad)MZ>b+48Ch94>d^J%@DswtWXW|MwHND`}8$
z;|b_rq03FZk_``w6UdtfGpR%BI1<M49sQ<z`0fIh1({&PdsUC78?1id*R6M1jz7zC
zXnr{!XStTxmq2F*K84l;+4wle8^%tI!^y+La5T>zD;%VR9Rbzg{oRoI<^1I`Pv+tR
z+n4D2gt6c}BO(7v74N*K8kSjJ#bJw0M5IdJa&Ck)oz0CW6&w0-&h%&~{9A-xyUiF+
z-~uo5_y@$z97%pVnh+)VAH4kg7M4AkNFRnQKrhJ<oH(pT-)VVbV%spZZS;cP>0hBq
zg2%|ZorK2e(m<a&yua)sPMH~tF03=MceW3G*$$-t`#<6N_MPO%&2%E~978(K5a`d$
z6ZTKpO0v(zk&|oFP;b=$zUy4Wb15-gSlR$CFihhek5ob6C`GDl>w$aIgIN#Cm}EC-
z(xwruSfDuxhW+&DuLWP(-_T_KlXU}}OB%qyy|>Wa{Q^IZ27-}=5Y3-%!6)<bgiQ@K
zsLU9D*Ty{O&R6DOPIUwxGOtAc1K)7N_iD7>p(XHF(V-%y_bd`sHQ?XP6`0fT4+QV>
zuqy!gqC82OI_d^zDLoM$9xB1-o*h^fb04(CRcQa)Gf?U`mOKAm0c7rb!?}&~iIG<U
zmu(Y9wtvz^q1QZ=>~+FngD70PZ#@1}Ne6wABn<jxMz@_V#Y<C|16$-Uzk3n~ZcAhE
zr9a^OCNHvk^J@&sUrF>5SK{;fxzr~AKR)4F9aL?b&HrM2h)dE^G|I~ylOlA;j=d*P
zIVg#1^47!jq7HU8%oDEB{lcwX;EBhd41$|x2C9!ypx^J$psdG5kJZG&3AaL+ZzV}T
z<n_R(bz(HvT$@V0jo>Er-^T~9&f=cze(X5gfz7fnxxEAITv|COuzGv~H7$oQ=-+0n
z-e*E5j}xMZ=3>FP>}>wx(?~K@s>K~r72-g4CRR*137IyVIA_2Ai1qoG=%as#<w%wh
zftMVWTByov>{y03U&&CP^9%XZf-_(`r<qGDlcy_pGuEfoOgc(o8lG$L0e_El*jlUu
z>yy9X%(J6tpWq)xE>K6g^l<F=O@SNpCSkqWcx*AK0#}O(<kjT!=y9C&!HV8lcJ@W%
zlU`r&+WinaeXGH4zZ2f>)F!&KH*pPV;kc&Cg=k)U&Gl5bL9h5j3|JEY6I^v@+nL{(
zs&7t}52av5@KLL#UB6&JrAc7Yaf%<=Rmy9AI9isrC<~)hl+nit{+}h~Gq3off6Yu1
z^!Oj&RZ;_6HXoxoVP2$uwgPQ7jpd|D^98M4GDv*9Fn;VP)Mhh~z6(4y&v^}1GG?IB
zQiez4j*_HhNAS?;GF&zDEqgDFrXM%Sl9s3J*yitwX`~+PMrI3#TywDT>ju~_+sA)8
zED5r{l2re4D?9JcKy{4;oZI(VI6u1zU+Zq8J|Exko7wz%CG+^)9iI;~G!{}1{u^lR
zuEw@Z7gGB78g}hhpnoiZ;J6G7*O`KqxySH6^AC^PZB6fXOVQ1lN+fq*pdgjqPd1!X
zCbo+c;8P&r_lifTFR_Iji`l{-xcr05Tw_W1PSYS{@jj>uTaFFe+qlr;W^5Cy#uQm4
z;u{R4`{qYnv-2$ZFY7EW&_9p0aT#3xZ8s9`@DW6!EvaS0EcA1$MD@?vY@YR!Up%WD
z-&~I;ic`mM2`(>$+^zYTs`*klF3=g31Yv@1dkUR<ZQz958E$5n2q|!ppzn{2AipQn
z2sb_s23x(ExVQmG+J@g?eL1>3b+RhgCaMhPde5NIwi{*t1rdweas0P^O7ul<0k6=*
zx}jmyP}*xOaZoVgZnIv1Uw|1%7+2uKXY%Ca_^Z%7W<So}mj~k>^<mZndzj+11bv2L
zVOPB+@oSn%wx+4j_ov?A{+u%~p?nt6OMQZ$wvI%Nx0^@^9Ov%Y8q;Twogm=95w!QV
z1lOkT0g`LXaO<4!koKUCk8?;Q#zS_T=0A>ivlN4#hoPwZu9(Zxio>YomSpTLPdZfo
zl3O*XgBG2(gzGquQT6jlORWb9(Yp_ypN|ALxj?Myydg+#5G7d;t*Oy_GrDF>I!3xo
zBhBo&|9&tT?RzWG$NV!GI@;i^wTR)l*Pu{77xX_)#?n<W<k0KUP?S*&Qu`(nCnItC
z?YR}@s;t=ATR?BjX7iw@X2kSCwNU)!FlwG#f#nU1(FF<oUco^+9K4)M7@Q&eE;AYE
z62>+vcm-$MmJ!?E)p&N3Bj1u>L*FsRh=;T#xpJxpSFudpF}HA{vY*h2&QnN;Se|8t
zj=i8?JqjiZY{)sy5G;yWiXv;*3D0U5Lh5oIns(TaG-*tSV}@htrXOYKvpfttLs{SL
zSr+F0-G>L2I<Vj6tib->Rd^Emlk2pcCy?pLfX|~^;cU+qeC2TlJ<rOMsv#YCmpzsD
zcXda@lQY2R)HH}KR>hO2rqJc?K6qe#B#!SoiSBFmqvE|%^j^!@(OJm?@q^Z=xb1_$
z_-Y>PcL^oqyu49XNexZ(uJAW4B8i;0Hmx~vimz07hciSC;93pqeq8?mkG>{=Y3d-B
zcW|V?%$QT3UV=Vq|A9`!VbJ)<+#hAP(e1DTIl8U_ja^s3hHI@*@URUBMA_$cTo^C7
z@{#>NLm0Eb7xzvM!^HCUoJg8i*|ZKdEYh=u+MN67??YjUjV50;uaPm)w&CvYlSsZ|
zCqB&EMXaYzV7mczI%d%;-ny)nOP#1o*6n<ZLy-w&d~iIR?2)56se8$K%}|man*lYi
zCc=hAnjp7EljgGB)SPLnu==(bp()yA)kk%D{o_hn74?P>FTYCcHnJJ=^gTi;qa(=2
z%b}+xV_{}bB?X2m^v;9=E@{#b7K)`p(Z^oF4mAV(WH%31?5N?Y=`IXZu%pSBMpEZ>
z893&?GZ}Ba8Xc{Ra8%huu4Cv5B=RZndzU7u*;0@58*-t&O^G@UediCxs3W~L0GEP0
zIPoQwe6~Y3xAuu19Z**$JqGM7>-H6$AI&2Zb4940W(gnZHl0>xq_ACw1g(&c;{T-u
zqMckcEUKGKFI=8LEnZ(0xC2LaZM}^mOOkN@P6;~Egt?Xumtf{j3tl&<7f-p|hc}x?
z(fSn+__a;5NQS&9?OsyE&-pW(25T<D;t0lCex1q%waaq0U!Gv)h6Kzz^@(};s!=-m
zHQbE-ivE&2IKpPzzPj5W0BoVRX9PajR)iv^m*7i#GM1LA5Wm1^?q}R|ym-KqtLUl&
zH=9&^5~{<!`&i7kN;ISTS8)tnxsp@A$a0+_H$c2z5;M0Q7P`qigk?ofF_d*J$8O8V
zr)*B7wN8Zklnx0>)N_c|h&%i~t4o}*SexL0WD37!Im_;*M-$WXuXs0BlPXP1K^Fa|
zq8S{Rl%E3KoEVm~<zQ!P8w9SffRXE!sZOg7bFJNg$TO1EGptn5u6mc>nUF(1&7{m*
z_5<E+NQ2Il3%uj2w|Jtv2k*?1Ay2CGi0_nBym5pnh(9qQefHb&?RS=e4W7=oMx4f#
z{|%xAFGh76j=+i`GwQrR0Y{n6pk|M%1((Fu!%VR^IAe<qeYNrvj+$}<rzs`l;;2m|
zqhumh4-UcVYu*@ji4dpyWH8<733iW9aGhh0ahEkT>EDnsWQW8tkYZe1{-YdZ%e})`
z*T1~?-qExp=Qw=qbAW7?nGrJ1sy2*AE2S9j+i7X?@}>y+rkal_{rP;n!3Zj<eHyL3
zvT@C<6|h#0u{7t6#y`>>U|Ole`6L(PQi~kSD&z4-_#6D?slZ<z)z7cb*Z{fD{^O06
z!?=xi&B*gLXRyt?5UYZ;ad}@0E|j~Aa}Fy*UG-$pe{vd|etzMOte!}BYfi<gJ4;aN
zD06aF3h}55%MbrlB2m*fVC9GNU~d&7tlw^iyJOa3|I`kgy(kP%d*9-=W~5;xi6-+!
zwNTWn6va<UVEEXZ#KY5^Ry;T$OnP_~;fw*w)aGM`Mjc$=oQkP$EvbxCF*YZUfbhC4
z7^9&@9(Y7>P6f^w(#!tcZar>Sm?3@X*@zX%{|WnNN%7-ZRzf9o7jFMCnglKQ0zXye
z;2-}3AfxF=ef*0s>xwE_b@2;i{xjz@W?#pLSrOnIzJ+X&i^AQ-GW45Q7k1S5@@)r0
z@RqC-9pB*w9=@Kif4namTeFV(Db~rb7)2^GUtz|=(R3iv35$g6cUq}O*XK7tHshCu
zh7F?DP#y}67vZ1qov7;*OC0r7N!-OG@;x>gB-#%{oN6p~_KYNowO7h3EW^qbK!Ys2
z^AIl0_=wul)wsLp9?!i>1b;JE5`KLU(>e{fh^!{m>hq?#n|9C|?hnNFWP_YwF&&Oi
z<|eL~jFHDi)89tpXovGm8nlr;W3(@0M&B?9kBHM^S7``Yxmj?c^)Sp$w8MKZpJUZd
zYfSY0g67BTF>*fhv|Hr!lj7LCV)bfnwwNqEAw7aFQUA%0nrK6WJFDPkh$1b^Ig68~
z-)Fn1J2+lq9B5Cd;&zXC!%x+<#&6?)F^5(WN<a*ryIR2e3)hk1{PE~_@+`(|p9S(|
zk9hN*_uRhsvNS0&4tJjrBL|Zf!#*=*YFzZ4?*}vf*}T!rdxXqSx)nRff28wP9Qr#w
zMp7;Uj~*T+wY4)*e5DyCm?v?PqZMiN%G;1@_J>oE-^(Q|d4i_-$>amup)~weq?2sU
z^Y^=gaPjym9NK<{6FjSh3CvYkydVS}-LG-}y%CUBuR(|19^tH^?Z{1d%IgWF$ni<q
zAj&)wwms5RiS_JsTki;VuiZ~7_Ws3(zeTCCP8RQ39!;iuRpGRJ5z>CBkej0&$6S8n
zShlVm#n`)cSXCDe^<)u`e~i(jw-ujs&E!ScS^m@hb67ewhHAEDkgUu0B;@Y{+*+?q
zsxSV8+0Sl(Psn_(CVoD-9+(T0;+}!YDtS<}p2tmC%$y;g6No%Jzy1<3w~OmA1PqFh
zihyrW{!a><pN|IUtwsF>E!@GpO58hB3EqFzqq4<nG`&|0CLKM;RrOgA;WduRPrAZ;
z=WXL<rTU=lv^&XGKLb%q)#$HMNpj5Hi|QAe@K&G1$xQn;OqhF8Sl%YYxNr@a?>dX-
zMTL_^^X{Oeax(d;%H|@c_6c<5M$l4Afnf1l<QCm7h4<&hNYUN3+#l5m)F+9DQK(Ek
zCpVYB9c%;#Z(qoi`-$^jpTuLiUqQ+>7~9u+lg284`X_H0#$VY4zVpLL0cT6MecMCi
ze;87OgpqW?usmUGK%To=$-g&^KnL5qXtoCM;F~@!bvTnZOf#W-Wv9}jOnrRRdK8V?
zu0ZSK>*(n{5iJw$K+B^jRMk`f-3d0rOP&9rd*)f*aojz@Z<D8(@F^eWf6U=B?`jH7
ztS_SO*Uf^a6&Lvahb4k;K{7cmoC2;JExD8Wv#8b2IL`Q*DE_#s%<Y~09h-f}616A<
zAJ0w931&;?@A-x)73*QpL5Aqhnnt4(tGH&af-{LchasbO2sD@LQ)|ub<j=82KG#c{
zdmk1|BImN4a)gXfs#uL%Ie=K8wV12@J(l|PYE%D`f84R-*RjZMG}+6zBS!nxsm{?^
zXkWA&`$;%SJZ?b32WqivmjT*cFX7ZIg&?B+zT8R9jhYvn#va$R_`oHCw9#24=XWqR
zj8US;mJ`zE`+zqin@AhW0?4^|(C(#Eh~)}X+LW4tIUdVF{>o$cE+0X2wMtml<Q3Xq
zm!XM~IoM?}f~d{S$F+BgAoq<Lj*u21%X24C_5M5DL$!&hu}c<ne{UrfL|^d!$}5OY
z8&9%=M}Ys+Nz|f*?I=%CzS1TV7bm9j*Mga!+ZT9?YkCk7c$?q0-Gr`ExR0uaQFy8*
z2lan$q?vXiXyWWbF0FYE73y;Yg<21U<|~_V=7D%Lnsyp{IVoDm$<ZCX2cfy;BJ>Qe
z;evyOTwgik*lbEgM~yP5Uo3%t6dmbQVFKG_usQepl|<ytUNSsbf<%5HS$9l^IEJ0Y
zfRHnM@vn03+2C4SnL7<U73OnKVr#+ch8+ppaE%DBgp;nKB_M3rg}Ya4(4JeDu)<4*
z^p{)|R7yzE5hLzlYT_5%ckT%q%q@UrE|IYJaSF5zCIi1fg4B!>r%n$jx4zvJ-%Vg$
ztTI2CaoUI`Ye<un)?-|Tmjwz^Wyz4J2}ykane|b3!Msb+@Mmf|#~lddde1#Yli4e2
zFPwt;nuoad8RE2GIRx+Jk0qJ<v+)t@QK@CG!Aqud*#0Y%lh>L?|2>by(g{hd^LPQ%
z48+OX6iI5=97jaX=vf)&Ci7h@WND^DICn?nHPUyB>7oNF<aWaxvNKecoGLgCnG02^
z!&+Y~-Fg7uTsn`Q7VEIpLWA723Wnf?+N6BeY@&Z}4H1TjFt4)>ZH)EBq-XW$-Tw_m
z`!+*Q!~zo7Y5;+!ZgP=tAAmuUBu*StfRAE#f~QR-pE-~~M4n`o&WJ1oJLX0B)Ab(u
zN@mh~|FlS-R~%GaZA8V%X?(|7J8+sg6^$y-5tqtryx>v~i;s^cRR+_j@#4v}{j3#u
z+;XO+HIE@D@B;p`4Cf!uoJQx&eTuP-*_h#G$GTZ1jMpJS)~zmMehoQLysIGGt@eW7
zw~fui*MEjvqz@eJJkhzQ4JUk1fn<{^Y)hX({m0+nGBTC%&sr-8%e#c$qLJj|xjYOx
zqe?phRLSA0g+%q+UtzhuDlxFIrJX|=u=V{FG;}RtUDdf%dHNKz^a&@a64!WHJv+3&
zsDv+UPZH5~AE;g50<liV(D8>DZGO})@a~Agw53nccWf_i)Eq%;Zs$R9Tp_-@J_kM<
zUW6-CZD3)1JnB`CA;-klK#uVhtQcMimgPod_KHhb@Z%@O887Ag?;OT1x{z9DFBEvb
z-Y-}-KAhMahvU0t5;Ws>7;%jb<to0ka|2fqOLj&OWg-SUb}k_O>+cF1Q!|B0_YJA{
zx@tZ<=?#jwEyWXumC?6b8fFUexq^NTYA08TvOm6Y2g=vu<$IU$;u!<_O*@vWs9-&s
zot50uf{}D|OC9d|odp-g<A}L!CkE*=4x7SA$S4_2<o<nOyWQWoK&l0AZj`2)`n}+=
zJsV>C`*GBsWVmK2Nelc&v;1d+@cwfN&~y=knC1qu@y{#V+x}G0IlC444f*&^ZzAF1
z-^0WgwWuAgNLL&gi}_>jLdP9lT)F8b?h`$L7Wqr4@7K}v(5NYN;rDj9C#Md@!>Uw6
zvyy8~HKAaA0K3U2l&n?5j)XeQ;C2hmHznYoRMxlS9MDDV2rjZNgvG%(@wpOWh1eDT
zt>p;Je(8#JMcKG^cL@%U)TeuIS)<>i67Ca8!RNQ?_^_ppsJFHeea|Kl(`qkVU+RbZ
zPCe%;d@aDw|2DX7eh&Y(rt^NS99?$yJnUQGL?&OY#frDj(EpDIADr_C3xdQ*-=-sA
z^}P@b7mX&3r$^A!Ut9QZ5yq%{l+6Q5?jX)k2Gfa$QS6f?X3Ths$JqVB!=)AM<s0~V
z)i#tbKZSaoXGy{fj;a}&p?CF0a`Jo(ihFOv<C~Y@>ak@|D8iVF-3k!Up-g|Q{0+X0
z2e3t<1Vuz5$>@Pv2$2z`n=JaFGhDg+*d}F4qeHOb&j+i<5XQJ}na3}h*@Ujc-Z1_k
zV}9?S#E*K%&XWPBxp4mwu4Uys)}s-E@{O^O)8~fOpQqDPmnXuAIzN~+O`R@F`wJac
z`%p#Oovb*pl2~tQg6mU<F;zl=hGgZVx3nQ@XF8ByOU9C$h9)#>aT<wA7(sPkzXuVe
z)Uvg!jp>PJ8F*dB83X$$m|a&TM-5Mt>otbpcf}LdViPKtoaR4O--RXZnNYf3ovPa{
zV>7*6^fXGgdJ@OZ5+ml2@Zr%^<XL1{&E_-|%+SNZ>`^50{XA&uFM)zZ4bZ8sDDZy0
znmV?b(F2za@#yjUxHndU1i5V{+2*WwmfOk^gBkpK9|h_>w1=N_@;76|*>K-R8qg*E
z9gG{G#J5&-LhVJMd;J@^h<OIIGwh|H&pUxT$ezPyQ*QA^w>z<FyEZ!SuElxFFXE?A
ziZ<!@vFGYUqHCszKP5~-awwB1bS0o6>*A}K=~9WOGW1%LKJDEb0G*=-IVcmQxpQ|j
z_qaGTYKxOr544)rGn%milc4dj3(lM|1kGEhV5Uwc_Gmxgx9RnwU2!8{x=#kW&z#|V
z_3q%>H}6q;d<pkx@=KIRz6k08{xCdl6a-Ziz=}UIblXP}Qg&+zU6Y)GRuAg%VuT&`
zh|VMN2es+&%n(v~=ND!_OG1lP{eq$cg&=kR7H7EN71o91@ryoO!~snm!g5I@N;ZSr
z7*UK34c7EJ^PqkUH6%xa??Mt|!Y8yn6I_eh$p^h#L9Fvulc84|$Uf28aQ@aDnt0Tb
zill20k=3g7=9_+0jnp8ok5%9^kfqPM&tRUV4oz6X+&#J3=zG(aXl5#qgorMFy00d<
zM=Q{gA$runTaxD1<iN|Sa!{4<qZ<q-k&e-NL`<%mOMb8iPlrE(U-dHJ&AJOl3+~{j
zS++1o@i3Wx^DO%1hT`pX8OV*l1vMcG)b>LX$8786#h1nC?L*C6_N+1FwLux^Z<s}+
zEEF-;R-cNiC2?v#ukf__BtAf46qUL%fW=Z0kXSg9CLc!7|8*8*CJdvV%n=av$rs-E
z{+_dFE9DCZzTqmR=cqIFHNV5#oN|vBQF>Vfe_4bIho!4<`};twh*2eeX7;q~)mluC
z-;KJj_Hh2Wk)(rqLx+qUy=>EpjkzC~llDK(?Jv;a>R2v5t{Pm!XMq21JHGmHE{aby
zqb}Z)kw2{r8J!X&*GURb+!%tG1Fd|LrwVx`nu`YyITMe%I84wyC(w}p4YH4;(QeoU
z_jgF5tfnk&pSctlEA-&qmyh6i?QY&?dJHzN>cciCPa1ingMU~rK<RZx)NmHZcmXf?
zkLyKf@xT-s^E96~zn+4J88^}II-#joPLi5SrZ}87kx<54e&QaD?dRvwb;q57kGg{H
zzXCvUbiBav@)(T1G72kJ9I)E<!JJ-mY!?3Te9T=sn+PeZbkWp4PtZ|TjiV-8U<$XM
z<!9&7a+gp(FxHh+zP97f+*(f(f8D|xlbAw!OgsjpZ6JmZyRf2e9Z~GO$KU7|z$;TH
zdWLzrx|g_9rMREaDY2IJu(@ysdeX{}5OTCklBWM`!1PSSTz3CC&vs!oy9$X^N-MP8
z(1JO$zX+ZjIEKroNI}8tP?R-4OGFyO!0jMoTR)plZm_;c|C(X2|E0v-6^@W=WK4q(
zI8%FrVXorY4XbhAqxlUI4xq44g)VhH&&e;3Bes@V(EVvEX;*TfF*7Lt?Rgu{ePEBK
zlOl=zk7xYy%f8SSG=|fZzs$KMD!|oWtQYmD7<^tmgdc?w__1byKj`#|Z~Z+L-o5GK
z?l?S$NCkFpQ(&E&$4?<~oF1K%nahi9E(I=2pBBlz6drxR?q%gI$S*VF-~GM~na|o7
z!$z7M-y)8|2Nq(D%~R~z+66x@T;o3X`jWxy3t(D$T+l1(LgtfV#vNLRjS{agcSAaV
zVm0tHPe*WHpSPf4AA2sj&x22QgqWk^2nSxjfOM&GWVOd{NU5C)T~=yD!N!nIxs=G;
z8yn)_!CD9e8$1`1i{%DGU}O@4lSZuJ>uxc2n{XN5W3EiJN)<_0)?sY-7RA@v+epl)
zGW>RBJ@Gfo6!4}|IP7RkS08t#ckf9M|A4s|a@&Av%Sh9zIt?0>xQujfkH%RyW|OAj
zFcLYO!TXlh!6?hUsA)Qhe7hq`6BnPyJ58)(H7r8?es<&JBWG|#hCIs`xnurWcSsC-
z&mE3kg#OhINLhZZ+mH2g{XRjktt@gMHt^+}G|1285+ts$6>H@G@Vf5P=p!2=tXTS-
zgB#h9qHRk^w+f69HzD(;X+t#2Bif|PlLuyuEw9<hhu)C|wI^Tk<i|Uxw9cJ3;NA#D
zXZw;$VHoL3pAPaR_T<NnNTQ$F1YZ;)z+Q1sI631DuD1VyIXNP9$;^vrA<+j*G?pOu
zvJ`Y1BCQ(KDXyJz8-7=^u9W6{P&v7f)T;f!M%#n<_JACnFJi+FS$klO`4$qW_k#Pm
z%bceDtpv?OH~GLRPk`vE^LC>gsMpUN+|nXT29D3c?KS>jKukzTbf552&qdg~bQWFF
z#n@k^{al(}0Qn_8lE{l1lcoiwctL#{PFiw|PyFo)OXI8g5z_Bq)kNk%WjQP9f(NLw
zM<0~G1jEGmpOCGy0pDre=Mr421fKN?Lh~cr$-M1qWajo3gj<`4cZEE$+i63T$E|{z
z@j^5n*9{Az)xguNTlk_k1F*x3{E(ST%Pm|%*>f(581)aQzv{<tRf<HX#DG4L+|2uY
z2@!0!l%NM>Owe}LY>YW_S9l{<j{5xmhk2*x(yOa;8HaxhmwMcU_wQ5yzc1{WCY4I$
z-YJnojA65nxPfP*8JC%|TA(}sD70_84U3*EhWS^+$ioqGw7oqB`a4YV11!OHS{ul*
zxD((Sp-C-_JGsoG?pT#;Ld(WA;G5{RSl2j@wn~ho^Ea7Oy?_>2aS!O-r#i$c@hIj`
zeT1?Hrt%%xVwk-qmfg3saPcc~s#5wIB!kwIjq%O!PO}albe#vc#tL{k?ha&`q~K&N
zEAYCdNus}*lk$;w(9dZF4j!ps%%~nz_e<gy{+I(BOSE~zKjJj3`WuXPkix%rmP5vc
zv&f01^Q|X>G5KHz9^I9NsbeyQ-&iJYSU-`7%0A{QW~>l~bZeJ~son>5{{$@i<W8lE
zcJspWGT7g#L2GVo;Pn_ENOjb6)P6bzzF(h+3qM8)yLEK1p#DET!>S&2)%rj+XeUna
zKZi!T;&kWIST00ClSt-Hr_$>Au%<Z;7d0Hm5|21Yv)zl^?|p!nXQ#P@WAmvWV;S$+
zy#{YboX40Qqv1`ND)k&W9mpF~5-`^uwKh#5iwkzZIW-XyA{EAessD?s>Xb-Bx)fZ#
zx(v3R-c3Y8gUQwO#biZ~32}*0A?}C&p_73ZrnDK*gF0v6^GX5og}+d?E05p7=I8Ey
zT5ML7M2=V}QeJ8Tl?aW;Bi5Z@7G?$4huWb2dn0oUr3){d-T^s5i5wqWhhkR`aH3j|
z@JaM)-bVd8h<<Xw|2ks%$BAD!sqKce#Z{38-8+M6(>3Us%(ocSypVp|lZ^Z-0qNS^
zjcbcs@WqmRG&p5|Vlk4C8d`;}*VkhEvgz2K#kv$r&Vz2tTdUZ~QQ$9fgEO1)4E3B9
zsbGf<jh(E=CoG*yPCdJZi(2+#UZ)8xlr*RDd$wTJ++#$yr%>R~z6xpskAl9%Dss+H
zpZ?pxI1VMQKt{@7|F%iQ+<`e)HFn{#|76Gw_VbASxQK>)<VfGpaxA{aal@M#gXZr{
zp_EJs-!iiiHKQ`1mc4habnGELxC`eVx25#J9P-XCgJihHLfOkhXv|NhZkI1%chnFb
zo#jbO;*`l3@7pkPUnF19@tX5nkq&Vi<58uFLXbowiW#TkHyK^l%{QWMwy*HFPc*xa
zO~ulrApWh^W9HYmjRkKUQG8z+e<DAWGw)mhvmUC@e@i9k*2fcxi09OsRoeC3drxC3
z`>T-ObJvCzoYW;&T6@S|s=_&mr-8n%6g{Cn89o1skT;^EXv#arzgrZ7FXkAM_V07(
z{9SPz_dSfC^lmb}m^B+q+1~J9fde`<oWU}&JD}In1;RzYabL|n2tO3ZNnHC4QZgsG
z$z|i1lSvvS#-HQ|IH1zfvD_Pv-<W^fk1M^q5GvHR3IBbs!^$@waN}}c2-dF0bTLJ)
zV=ZI5pG}5)wrj~$^Hua=dKxcf(TYW@{7A0lIntCJgeN!K5zo6H1ex!a;l?CGYQuP*
z)5~~hc9!Em*B9}o^;a27NJhBFqo4Db?G6%muJgK`a=0cuo9CvLLDbdx`2AsvFkUwg
zoPQle*9s@P_4j7_eWE%^*Ab;|_s?_o1INjtU1Lz?%Sb4;%HwAT8j;p@iEwU=7HWp~
z;{FlGaL+zBIyo;L-PT0lh~NEq&H5VGdGQhVWa@0bWNb9??DQ5iCQsupm%ql?*@2)q
zc#Qa8Uxej--B`7TaZLtx@CWV2VTJTZt6QJg?9-?SuRIw~(#`*Y%WQpmZ}|*Ls^h`*
zY##SYeG5&VQpZnO!qE+$=HPRv5YFE<r%&}$c;`(vH0X{2N$~u~sRei9z;h|GdR0DD
zy;h_ze8Y+LX(bw<^9K7XJ8`42I@T@x29y5m<aRNy*Kb)d;%mhGvnJ_aHS;xy1iZ5J
zbU6vC$vI?gNDAbKZo^|w9U<`7F5>ojA(h=E1Y_5Yq|I^+S)FwhH*7u!8&)L|^Wl*&
z);Su|uQozV$NxDhU-5Nw1eVJj#*LvO82foA)T%MR^CU;+*^D5$3wt<awMcfaw}Yys
zqw&~5Crr_a#5b@HD<;=l4H}I=J+^23GwnNs$7W*GsAm35dI017*$XY2-=Y0TY2spc
z6T4nKL_?!_(2@~~4?MNd#CHbc?Qo<u)CSX-f4optmZay3(=?9`zT-y{+8i{5z{oB1
zd2BQn*BHTv@Ne;OejS_%`OW>cGX?MF4*p5vH}38#cPwrkOH$1&u<a&ua3_qV2N~-$
z;Xo8bG!*jwqNh=GgBmr8%pf<MlQHC@fGU-}=iK>vaPI5CZ11rc=qF2j<WG>nH!aYi
zGz}M<CWEGG7R2P8f#=MFl{duvIKM{Y>-E}n*^_$E8GQl%H6CS|^hw0uN}h(-sG)Jx
zZ{Gfy0+{N*6m&W>#?O7$?f=1g)}KPb@Rk^@RD6r27ZZuAQX+{_F~#DvpRhmZB8Dk;
z;pVGR=q)ya1iE#>3NZ&7b!ryb`y(6E%}3JiPp{GOa~~GvX;R4GvG0;09}-o_|5q6T
zHG|y{ul*F<r*0tA7%K!$w4y}eCoC#_%{50ifvJ`#IK8nV*VgWY2N~6vvrZBAZ}!1?
z4We{~q$)Ag_91(ITac=Y)!e<&=lE5jS-Ae<V$M9<0q@~$&>uSjDrl9!Fe(iT>*j#L
znWJd>b02Te-3ej=($s0k9aQaNoXV9eiL8|pV-G}Oj?p(%msbR1u?R9&u@YXLm8Ryu
zJeYUT0w-x106*V^*S{G7wZ{}GeD)+NS$e#W@*j-cxu2+bu`|G)@7P*TK_bS7dF#DN
z|AX_qRc$sGAu0#w<jT?T;WpCJ;7SITR$}R}1=T-zl282`N;LDT;ZM*Sp_Z5yra5ot
zQap@E=$vSb>`ue9f*jtJ%||xy>eP7i6^xTFg#zn7OxpAp9*Ol~?u<^nCifc~>kpIi
ztU<m^pJ41hQR?KEicvGX;D7@2eyMxI)2i3_?6ED?Pn(J}MjGKj=3)5FcC0D)*O3zr
z$uL3Z5jvmF!KYr@So@zQZJt=n1zuqum01dOcYqx2`f?OHXRJq)^M0hgR+>ar>+<An
z2~6-mgk>HV!K>gLig;Aup>++o!tfBe`f(lq7cRgbr*D`l;|0Y+bv&k?#H~2{47rdz
zEZ!+eivk}CcI-E%T~n?g|4$Yl4g&SKJRQ<nWr?|a9$w#hA9ELUmz(^SA@(CRv8e8n
zu;=UruE<asS6)|yF<&a-ZJQn`^0G&n#gF+rveW6gfL;9eeo3OiGVqIq6UfG}LB4SN
zCYV^PL0hXl$b|J%@aR`ZmbdB03oHI&DXY)xZU_>jH9z6UJhmt2*IJVVe<7cn69=Cj
zY{kWfzVPqjI2!ppl%GC%J9%qz7cbIgRH~Z=FG_A;=FlI3?z1L=zo(+r6S4nDV}uDw
z)3s)^^iRU3ZKvS;^C+YbR#Cn2mgpOQ8fAYu(!V||OPk5&XMan}uQo<QaNKja%eY&1
z4U4%QvFmXhJ16gE%wfek6;g7lhWjV6l(gSu%$Jfg5atmM0eKB*Q?nAjv$>!N`&pIJ
zJS=-UgH*P=qM6MUnptujV%Yw9(#VtOADYHKA8)yg<0*JGY9w_p=!TYCVYnjJfCm0t
z&mUeknO6`s;fIQ|;oiv`C_h_=Y`$qkd}F#%J^eGglguWbuYOofYgVMk@=N&F=L*2=
zRu?*@XL8p&UZKZ-%lQ?iJz)GxhA2GUkHaes;7PazE@JP)n)r4ISlq&=ZP10{+QlSy
zhdf<4%9w<7RH92#0hTX$#a}O*OA2GNxNy-&+)!*OC)a8~UEANIerhrQ@8b}}U0Fe{
z&@|LxXWg*onGkS61LyDh4~yGo&|d8l)Nzxgt%uXmY0U^abD<RZpz;(wRqhK@Z+LUd
zQ~L1ToUJ%%uN7u(G9b+{V`<PPe>hO)g^4qoz)Hl5>kAx72jjKy+O2t9Li=W1Q@suE
zKMuvTxwC0Z%3W-CaiBihvb?R(4Mj#QB+X0K!iv93(bVN2s6YNKT)6x_%$lExWc)rX
z&OC-8>zAX&Pg(j>dl?<MO9onI@_hTZF`VcA?@(G{4&KS<xojoo962RUTYY{o=FoA{
zdsG~3s$^(%PYg^JhGF>fLhu{=0n=Ak199JnNsB5_ZD%~ryAX->;l}Xj<wDTco=&XG
zUFlCq#OhlwvFffSREU2U{`#5+?_5Ob>4U5jvg030oo(aqKaT=?=VbUhU7Q@<ej1-h
zKIhV`9H@Qi1ZW^pI4Mkm@=(pb_ghx$JKy2J`aHo1w_{{LaRJ+nogx?~guwGAOcrZR
zeotPAKKukO!Z)47{MQGjdzHz4%_RKnGl9Nd?TEVvC7@Q~hG6EQX;ju$5?9(zAm1-c
zrRUOEZ*LiMo3*T_uG7`=i0UsKGh+{_S<;BphwgFhcNS6Ik~n_+G(&pU?JXL&7lQ?{
z<WofU!Ps@#f;iTzqb&FBd_ajFbsi5pXYPm86X7T|p#!h)(Zbs<y}0QBvhUrS^jI12
zJr`1Nm(o~VK3AVsBtF7F;^TOOf{B=E@feonS<+hV!ziC52L2Jl+#K0N_WAk%ZF_Hk
z-YW~zIIKZiT$q!rqYszrFs9bH$;7+(2xMPyLAB{B^h>4~`DPczU12P?3h8TTa*{D2
zlw%<MPa+p$ECS7A)u=<-Io?bAAll4I;toWe#b*7>e4pM&43e5na?c;)cRGwlv+)D)
zZlE1*Z;eH#olX3OEhE6M!x)2t%Q@Q%IavBSlq_62%sH)2V@@(@x+f(IrJ^qL;@68%
z(`_QvP<_TVj4-6@N|uuQF(c_aKi1ir{D&7?q6K510#7hDYV(Do!lDQXxNBJpVAKqn
zl~Ne1eF-kje#2MjhM})*JkGUwjESy)a7fdVB)VJUr6uRlQA|Uq5;2u{MyYVx0Y<E9
zrbI6&D$<OjB_N&n7Ka{+QjLB0xD+c%IJ0>_Nk}+T&i&M;wKeHn+r52wAaXTTc4FP{
zFiqP2PMdCTW6YoRQ>c`m4cY7bgr7A|6r<*D!nOm!klt|!%dg6ysOT+pn4=8E&u;M^
zf!jc6Iga|z+eL=H{Sw+Vxw1^LT>0%!8~OLTBZzux7T>8U!}$ly;7gu1;-7(M_<|VF
zijhHrpGQAq$lP_pK8-SbnNfv`j1#BzsRmOn{=uv?)^~mO3QaxkgT%)M?Cq6?eK|HH
zGprQLx4z?~c66X*L^$yg93ub5$&i0HM^f0MO2bPj-&vDbro8S5Y|Lh!w8S`){3;SU
zR((WI-*`|pkRb0i%hPKcdG1n@1vztj43%Iz*W9LTPWXH!`hRfbwlMZ4SEbEKc-Rxu
z{ql6x3W_0rTFYh%H;}XfpmiS?z=D+H{JFKRc&%|KuD&v#M$>cn>(n!teovehY><I<
z>teYh^Y+0;tsNMkr%2>atjGSsK(0G05lW|Y!qBhtyyvKwR{ya*R*YOLJZc_A-3;4t
zMz;g5A1MZkTUH4>$Mf9957ty%Dw3aJTMsY&3o&`#Z|G|@qUIKoAd=lINFJBPKf0$z
zoJ<~ZjXz#<Emxb_j(-!{43=?2zgforg*lm}mcl0#De(&XviZ)Qso>WThySvI5R*QF
z_^v@zfg&Qw&YB6Ae_DOJy#$6Czk0RLX|QGcqc5NSqo&q#X?ER8;$qSPesk66kxQTP
zL-<n^FMWiwD{tfHWxD*krR!iDX7kOr#tL0uuSC)FAt=743Ofz+d4uFhIRAePoe5h`
zT@;3!rAb8tlB7r_l}dWg+A2vxB_tt9ri2hOhvrhLBq=3nk_KOq-o3VjBq1aTDU>8B
zLM43XA9P*4=bXLQdY=0pJD*%Fn}+F)tA$hfS)WLI85eXpK_Hh|PT!6GM>p9|W~m8t
zL2>y>?8h`(sPYJA@H^q4@a0yo)-Hn(H!<M8@aK=M8K_ZFPvyO?@Q%|ToR<0?L9&()
z?03;3%N%mJDdN^-lt%}+y_V<u2o8M5D}+qd8G&Z;4}~2k8!-8n1@1}b&lNnsr9k8p
z#Ew)ZU1z^?aq6SEFEeJMC%+#YyjWv(mj5>UuIl5|c;1qgR6J;1Hx&Bm<#0a@N0U-B
z1zIT-;r@7L(h;qNH1V_xx?Re{;t~sN&F!Osm3_Ezx*Yo5&f`j~tjI4bRkmnJ7;(sq
z!rsa}%rDVr&j$0+&)tvfu-Br6d^f&X=tKkmCZN}$FuH3_GMzp7HvX}YW{<D6;iNBi
zaPtd?vs=RPZQNp%E4fY$_D*6orhyoDGz(gPgu}m3A%0O%MBJyuGS%F8H>*7B`mMo+
zg-c1~2M4_ThOiCGifGGocV_K8AFkiHhEEnwATRSj(61Sc-WMhy=YI!>54TZK$+65U
zz(`=}U4|M>V^K%h61C?Ev2~0C^eO>y=PJ=uw;WB+C-IIzK7XfJ3um&o;_Cmh=mGUz
zknvuUGaeCzmXlnVUdb8yIA;esp4~>8=Dp%QFaPLgzU%P9K@<&LU(lm}l-X)-BWC?D
z!8LoeNQ9v~6FnwN0w*59L3f@hvxK6~tCf&bF3U~^d$6^I53zpKPVV`^AskdQBHErg
zXfNIe;M7K4hT=fKL;|B+`R>wYH=Mj=1Y1`(7Y=XNC7mj-!P|HgdA?&JQ_s+*VzeI#
zpYgB$_8e`RpM%%&vwZeWhBPc*Ec|XTfdwvjz{R9F;k`&Zpoc@rs4hQLy;)87u8)Ny
zm6xD+xqubgG|}KOd_S_i0OYSLkgGBqNO*ZUIJD0a^kj`D!O1hQ;8zkqchAQ~)QQy(
z?-GPw<a4_%GwI2T-SEx36x|*d;`b-2khkzE-m6)L^@jxb{K`k%=c0*@pC%C(7jJ>0
z*8@CSQ;5%=7IIH{W=OZ;9%A($FsqC>vPMIMco`^x?~ZTicjW?>i=KkSrRg+QX%4wi
zC<4mAGobicDJG|CVMI|R=JS3FxdpcH?Vtg!QAozr&zDfR&5$+r^>Ev4HsHIBXX!xQ
zNER=gPh|HtgYnFGjQI{g)i(;P6AU4XJ^|@`Wg0s&54Y|61*-ZF!AtZO%5;olZ_W<m
zaBDPo_2FCEJK2${ewHW2kNEz{s~mb%bPrZE4B*(e(o8=q2E4ER17TYj#1EeXJ@d;@
z-d;>MUztNM$~d9H+vlh^=_E*Jou{U|Rmp_U@kGK#42N%?!mm+hv8P6h1;#RA<%BZM
z-_w^Q@qTTYS|$FSeitRae?)8kw_qX0(N1!W=8_NGkn$@UsCNu{S5{%&rw6DWyORvW
z@DBKvXM$&rJ(xNn5qmEmfuNQh6;lGzVXa#l+<kNz9~q_My|0tmNC@N1WJfXYk9#mq
zVXW}y+&@^UzXCO;OS9H>OW5BlL1@Q2@8q1a@zz)mRMDRdDW{{sg3q>kn{6R_rz&8!
zp(zGDZ3d&jRhZH+ihXUq$^COV4mD?aS3BP`6PdM(*f#(kl{p3N)kkra4S!x%e*~xQ
zUE(IZJcaKxM3{BK7_0Mx-%!!=3BG^pM0Us&&~HcN*+jnp%#~DOhvd`IW6pGTC{>(n
zzA;|l(x$;BUBAvHrr(Bib0acx74P+YtIJya=diM^i%DF{K7mWfadN*!mGr)9LJfKg
z#2f0ta`70>)b}|(p5cx8x4zL*+j&$|<|b;H{zYY%O?cAcIUZiCMn;6?@MqMQ+(hqt
z)Iwep5*zhMiLwfLozTdMv`)k0ojw>eqZ^7dp2LTE-el>bwW#-)XJAVwVX?wJPW7)M
zS;_Nst+o94%+UbvW4exw`m4x$at@A(UB}rjv*D2Y460%%V9ES!J2cvY<aH_V{0B)E
z`hxez`RTG^^DOH4cp`VULmVIJ?Zu$c>$$z#_hV2~)s0g5H0%-QeVPp$c=(P4TOcP6
zEV>=1?Jxl;Ee`EZOh?yOp?E2*877JF*`cHij2SD(()N#LUDcyWKzs~M`_F-NY52k2
zUzV)9trCNEG+Aca8LGWi2_5%F!=jhA&>kaC4$ie?o=UN}c3}}p*+>x0Tbt;WV;RU7
z2)Kft@#y%=9TFPH;zrL~{PU?o)WKCqRV%=Cu_83wnZ?=j`^x;Wy=Xt-4P3fBpMK6V
zXYJ>vv$F}|Ad$O?40zrXh=hA_DTijWU4C=1zPtkL4~wun8#*}gi>EPr-XL^anzLUg
zWY`DMF`$&Aj=JSmX!rUEbV<MC7R7BR<q-jNs2~VbTlZl|L=^s(b6^hNeE>iAb1wT|
z!@&9JM67H!C@%g<+dpd%t4MF$JuE;Ib6d7(eJDo8_Tj?dC6Lyyz@{W^z%}#Q=u%B(
zR`Y8u_j5F#v$lUoTbCLMhjK*O->#RCv`|2_dEVjYyM3V6vkRs?=2^k7j&b&hB_Q0W
zi33+Ip=-|;+<fN^v`LAwq0<?#{~*#;0cXgQX5P;<S%-85e&F`*$iw+wF=XlL=~%7A
z58O+S;aoUDo?WelZI8pSI;RuZhX=sh0I90EG*d74=T1lDqxJ`iwY?*7O{WgJi0-Ck
zZRer&WhC^*9~0aUT}Z~TaO5t?FrCMTxu*ZjneU;K;JVoaO!MmC(nmc|UzRTvO<IA;
z;f3Vpp*W&iHh?OY_fUOJG7j%6Ma2)c!k4>ca7<Y!)`*_K-yubK_gp)tGu{iAx4gyT
zeJcf)=M7jo&yjsRpv-z##}mCHg*bFB0sQ|t!_|*t*l_SF&VIo>9O@Lq*(Y3xeZ4ZN
z{dp3N_r+n=tXV|*L^eEI)`eyo(WK3MISK3ZhU#bf7%#UKN<K3%KKq^a1!j{CpK>_w
z`V(YOz|%`Fb<pLK3FP?Cxh%?T97}#kY00x**e)Lkxf7eH;*DE0SGF8h%df;TGJ@4S
zE~OCY2=5<EAuFndbYa3Z?)3<BR;?{bA9WPM*27wuF(^fwcU(Zp@Nq2D{wAkqlO!z1
zDEONy&)nB$k>1_!QF`MP;`QS$cUfW;^!F~s)+_CFE!T`;Gn~oNEtYKfx(Eim3?n)d
zd7t{l-!Nul8CYkSv#(=DVur7T@b!%r9A{y|qWK<{{IPqOXCH~(SAWySG@c9lC7u-Q
z4n{G{TsRV>MvS9JGe<sCCpUZuwtilM<vQ1d_1nfW9p^eYG~|ur#PhjKQ@%5A&)*Ln
z71;VA8zS#Ah*fSc;K_B~RVfw=zHe57x33#>ix)-vnFhcnG|<ix3U6`)aB*2VkVE@G
zPezXA%B>=&D)rDeekQXvyJ~gAWHt8MC(ya86xq)b88p=}hg+6D%xTs*vhmVYbeTAb
zMNSYSvt;9N-=-|EFPev=FP;TC=QFs<>>!#9DUn6oNMaB>pH1L*-&Io6F#G3dvf<@5
zu-f?v^h`@RU$qoST6F^L&KC-DBfn$D4L7JvlYpm}9^<dkSvYjf3jY0bgA>iq@p^R=
zwNvN!hHfL+2~A(zaKi;gZ#Q9I?9I@4nG$mmmtv};vZ?f(SfcVQow_xTB(8@ld6%#d
z{kHPFU5O7kGP?=}B)idT^=PV-{ha!I-NW4tHDvoYRHI_XB|%h)8(JD}p{@Sjf<xoa
z(xnc|Sf+0@d65!AY$pgg_Z5lwc@po!8<t@beRjB)JB+s;g`iklAURr?1hcDE*^BKD
zFs`juutCk0_=o>Qt=-e;Id3&~X3}c%=;b8X^Pcw>gtQ3Vl`<f;Wf{vo8^M)YnNa^s
zEuqT#3LLqo4J&?`;fZ=a|66H=(f@^EYyULrr@w?-@0dVWuAa;MZ_MNtimbqxRq-U}
z&R<j^^6bu#KFfG-06}{ltcI^=2n`fdP}L~}16N3}l&;@+I;0P!>n-6mzdO3HMH(IN
zo`H$q`R>#tF=*#K23leZN#*cARJmS@9bpN`_^jF|zbMSAsDSgQmB?Y03n=f_1<q1-
z_<rCn-Tco>P^7h-T&X-tPOGdS!id+fs7IGpmM`RN>r%-5+4u1Lqg%M9<ORCaYT<){
z=TPgJ2(#6B|MewtqIbHTb}Zywn={-ApS=*A+jR>S{W{PkI9HGwv7IQY%d#o|T|wa~
zDSlTi1%ZR-1?=iePGQd$a_wjkUXI}ZPXFWZ%k&}yB_BtJ&U?6=&yxs51M$wR5p461
zuNdo{fj?z{a~1ewhRRF2<fs_;XZ#78IZz2k6O!rUw>GFeK@}6cHxRGqigZ)^2|Q$W
zofb;-UFlt);N!DtXtrdShQDpVQ|=6pE(EAZDTjd!o*|wzk_~4?aI+haVS(Ql;oSWj
zsl~wv40x@{rp_Kod>7esj@_k@@$8VW-efwdZ(GL6q_^OU)+Vf|AYA&y{TO;u3|Fh2
z<o8p`#3gSAYn)yKy%kY-lK<UOmH#;36WhpV+wnwZX$4L%Q)0uS(u7#c5Y1_4(Ro0Z
zY)k4QoYrB^|HKt!D9y}r@`(Rr-n&<711<(a%I$YYd6OjEm-v{gt^bKZVw(hxWo=xh
zLN%UPEyt#uG{y5TP1tSaPHeY2h2+TwrXZVw*AH2+FYcpojOh)*-z9pCIhUZzP;!OV
z{4Pw?H-L`3iDb-|Z(RM3#oYbCS6DFA$8{f^#`@HL!B^|?WaYLCnEipjd-8Mg#&Z*x
zUHTv`TCt1BnyBG|tS9()vOKrMeH5EtcpPtj34qScIaJi;FKUWs;eq|xsI9#S?04;i
zdneOi+pknuwl;;EbS;mo+&dl4n_R<eyHPAde>%DNt^_1QZsO+E26W55mvC@wBqsmJ
z<=@87(5$71E=!|mb$K0>dQTv({9eQ`VhZo#O~->XYoLeUC50X~VTruECrV-zvAKT|
zO2;LE>!M~T)tScJt7M5{n<o9~KAPN^A<icGe8BF&bivWaZJ_3&%k0L=kiu<===?zf
z=7cRFwU-94OnV%@*<KBqix%M29sK;)`8$;fO(Y^yv$5ag6Ro+pjXi&LkR+^nh$1ob
z1S#fg@q}y*?zw1%RR_&sTHH;V+qi``7gBy-E5nsv{w9o!(<VoxCNMMEg<NyMKkVK5
z8mqIm5uH9+w%3He6Q0=$?@Fp^{`z6s#y$XBK9il#^W;4Rtr!#~56AiLj?%3~%<va~
zM+(?UAAh|_-WBmZpVQhnekcL@KVL>gZVgUZ_LOqop6qGuEc}yg$Yh=^!(>Md);-f4
zi*_v{!cno<JmLy`H=j(xqWnm}#Ww8l$_7tcJ3PMZGJ4Hu$Bg9Lg7KEiKyvs!M23II
zW~oRl4w}jNM883;%HzU2Q6ilC>=eP2>CaGov=Lh%qek|Ren#(S@ZEQ%i*R|-N(^_t
zf}(fdpj)aPN(jz_efM1~{OCh67q2GUo{q+zPkcVHbqqUmuNJ3u6hPWZ31VU$flafD
z&~AbkIsJAH1kATZ5vjwp{OB=(<oU%QvOkWiy&H)l7SdE9TOOWXJBcet+=7Ttr!dGv
zv*Oh`F=*uX0UjO4xS(nB6+TTq=*IJ*rd%4&KF*qjey>ka1>H}OHICwGA1ijZCK^Il
z$HKRmQ#dqw3)#lAoXNF8y7|yJ&i%dwOiHN2!&HM1yBsbfZn5yl=Od8PT#U5f1wBxb
zM}_lxuyU<i#dq<UXnSi7$$pweo*WB;T=RUY{dEBm$rrQoubnN3cqqf3DlA~tO_kjA
z#wkpEXfjAUd=lo{?&ZSMOSp|<{G9b3M`KYPg6|xsvF{PH`Tcyjt{EHCr@}S|3u){s
zUCgQSBK;R8vIA1(cz4%9*hux6>|-_7Te+MCrpeQ99%Gp4?oM1Vs7CIYSde9AA;RJ}
z{kZhl4E9cH3NBO~$vkfIJgqJ2!qp4r!rpl!&^pkU8}rUg=rY%g1=a`B#uRN9d^;7Y
zN8dvI$r;40$P;Uq?jd)jkFbW3BD_<rk1KUM$7$^Thoda*pkV{=8`B&|`i}6f%IIi#
z<eN+tpSxh-*0ngHNt^}D{Y3A*nZ+(Ne87IQ2rPQ+iYc2fBlf9cMaD!nFr^Ch(%Zn2
zt)vq2JHYx;p3tUY3EngMjES3yxqX5*p~;aM<V=|#kuk}lGYgwBUfNKQG5ueKo~{^X
zjb-4qCLCLA)yU4CWe8h4xPnd3=|Argth_)5gev8nir0GZn<;~$M;GG7jPdMV<aYkq
zkHB{^-*MPOnsq)J!z@w_h_Avh2R7lj#Xbu*OBBHChE(i6sLd+d9#nWuolf`W-Gog(
zsW^F&7}z$S!JgS>#MC(eS1;8g?%C(K-$`MlRz;NgRE9xWtrD3pTLX9gJ;98_VuIkY
z+wkC{pBQtMcX#@~<y<~Igc;IbV6(go4I9501A0W+D~}W4a_kNbwD%!NO3H9Un2%dl
z&Lw`QHV~b0dZd8b5s$$PE-Qp%o28G?H0}+yYxm>jZ+V<e2!~~%uh7sr8ZFK4p-9IB
z?)FbJmbB84DQLQobAzF{$Z01j7_ZE(-IzocFODFvS(!~;D2`7zq@i+_BGH=n#mZpz
zJp3<x9toNvPFfXD2nL2P)6He~F)oDf(ytkXExSV@-@=ToD(%G8XAZ-TQ&B)Kw&Pm^
z1@@J9Z+^QnA5|@;Gv^va;|L27{V2&wbQNih*iZa5wjb5KUh?}CHMYU49Q}7y3Ae4i
zjIRuW@#^-Ac&#)FZ}B~XFYCYK(crmEcj*8Z9LKoN*)g2rOhvl4`T(velLYCTSwb)G
zSU91tL6rNyqweTG7<fI5n$9og)Ne1PA6CiG;Mj5at2PIt723J5rb==Lc}~MqNw!7C
z8@x<nAaLqEh&g^4haO8o>G>(-*ViWar6@}NwN+rykW$6cyot=~S231dD#g0DS77(G
zGuZI>F}5z74Qh8=aXApSrmG(Vc2@Ddmn7<9G=dwibe&s2Mi$Jb>C$_-n#{EMG#T#F
zX6?V@sDy=(u1)1R0Q;5`sUbD8BS{Zb6(ewa%uF;h;vE(FzrpI>0xDng7gbkFvcsE4
zv--6QsPE1~GzqT40|GDRHp_<{`#hIf$GwKS$MYB`Qb{8`ld<xQAz2-Jf%Mp4h1JR{
zupvMK;+*CPj?2lihy@a4rqxK&AHM|K<}_om8ix%>QlU}I4d)n*WQJ;kf&;Jq@~ngr
zsIcS_ikMEfa&kIFreBzdE}vF$bTAgLiKXN8S&A&-%vGYISB&#^e1YG5H+tKP)7X_*
z2`(!$gf-`+h+zFpygN_@1-nf_>-7jWAZ0+E)}2E8ybrM3wU(1tl43Wz(}^iP0uwGK
zz>I4m5T3JyBq%7cwoJa$k~<zQ7V!Ufevc2U{&DvC!Dzo&mW&@fhy6|yh=0LxoM?Ru
z>Td>dt^S8`*|9M6jFF|icXtvtIS%}0D6*A@Mu1P|b5y^%m#nj|AZP!&qs5*wR2Qj%
zF<}SDXMX3|&SxGjEmLRl>7!W6hZ}HWzCPL+U**oVDzOf=M$C;0rE3bOVubN5K2z?C
z87<bVgUrWKPG>OpKmNCMmG6<{*`S-jNqnW61;fkdLw&<N;ACH-aQ6WGeKnKWy!`=1
z`ZmxJR}CGJ6}b5FPSjE9;QSo<`QfVuVZPy4Sg4wUPkB~I_V#(K?))vVJlaRC`MkU0
z$Qpr-;TFiwP$lS~!t}4L!=W1vpwn85n$BaGQ{+Zscv6CiFPIJKd&QV;L^>p1H-b9X
zHni4#$(`Ex5tXd?cVoQ(kLmJ!7k(#F=-|T2oXp5PpP3jr>_Af0Us9ECQP?FS08yOD
zj`V4v4S$zSG04NICi0}ylh1D2jOOB!+J(mlb+PmO1z7l29rVhMV{EJ<J|61AwxLAw
z+wC|>m1&_6F&n;FO=Jz-AL%1l1*=vsLATAy#P9i7lGe`Wx>CiMEsi4|{Yt1Bbsdcl
zTm@;JPhh9M0-Cpd=2{=!;@o(5+0L;ASl_AvN5TYTu9F2$e>lKd<@jU&m6_bSrTWZT
zu2|4rIvJloS;Iap<k_f8v*4HZVp7(xijQj=u{C)VA?b>2xZRk`@2udvBr-%%Bnmv1
zX|vNCUZLZcXYki}7P(p#g)d{<aBtUT9QF<YNwGY9{#Xn46}7^?%sSNj@3_EvmMj%Z
zxea|W&e$>O5OlWIL8|&I9OV{5Tmt(9;pWCPV4(-e;&V+0%FVI3_XD^c$iabIU*PU;
zJu>G(I(W6HGVbh1RxI%6l!SxuED2EKYaaUV2Ug*eiDrH_EI`|iP2}fL@3$W!-Er;|
z21hY}{TQ<GNeao?Af%P6{|P-C=A)zU26o?Ef~?#F#IPU$-97+jj>!iveopQCR|gmE
zJ<t6O8V~OdrI1<|dGe@hfF2lggm@_Hb9X#Ea6n@%Dw?zkY#;03{Bd&ZSW_Yz^|jG2
z39?M9c`|qXwmh8JD8|IK4$%y28MePgob*N^F3&0iGm9%=eE+6!(d1nqw<w;j{t!+T
zGhJxuL>qQ!**!dTKAH>4B|`5uUAkh7Exh=Yjh6eeaavO-loX7_S6g~%{TmOg->(Kv
z3-3Z%2M6`5J_rpSPGcRbcwbV`c1+DzAydY;qt}{6lw15A{aj|?u#YNxoUn|BiM!H7
z^Mx$_*<J9PRe@i~CeYrwn0(d};O-ms7~c>KXFfz@(m5$qO+Sq(LEc1Xo*d0ken@@y
zt;A^6E8OA)1?E3-7^BrcbK|R?aeJ?t5#0^ZXf#KL==&GLmipPK%X`@ax_)rd^Dh#a
zJJZPtBMah`dK@MG)RN>$5$OD`8~w$<!jG>v!F_EW-PQINhnkA;z_orVGRogd%W)>^
z7rmr+9cH8d?nX)<Dw3g_%dmIKbowEa&s*@{;Y!5^I52+__maDf32yIDf3+m>JK2bt
zM&)!?c@~~1euu^q=aAd_mx~nT`==j8p<+GXjg~WFiv&_E^H>+OI4LrxyL^}Dg{I)x
z*azIc@<$ly84M+Zxp+W-BT+c!k8ylgZ<OV8o`*3T$5$^#CE+aMXMc(_j%`7a(N5f>
z>DgFiejTje-oXsR6i!Dl6Li)hHEcRf-keuu%d*C!Z=EL7DS3mrK}8@4;2n|$Ji{ti
z7Y4Uoprgkj`o}q9-02bUVQUbc<n!=eE;&S2h37PUT#wf7*94sj9-M^pb#9iq1zBQy
z3_lJUv*3AhZ1E2fR{N6j->!R5nl>52v{#Vwv`RdDaT=`bHbP0F#qy0evh2SB=vt5n
z`W8vtpt~q0)EZ*Tv#;>gUzuet98X96`oTMs9oUjxs_fm-sZcFj0(N`na2p!ch@-?h
z&g9b=Rv5{1c)pzjZc!CVDf7K|yQie{7NhFZnz1xcf(=ia#yK7|L`M+?w&sg9{>xHe
zkEckoUVe6BZTm*xcd;4P-jZO}M!d(X%8v#KESc*?o?9U&Pu4Fvg?1^bFlKiIH}7;5
z?_b`ETh3nK=C#dX$Im~-C>g$cF+mFV=01koBO*|&eG@j^OeEKPL^0HPBo=*G2wNXK
zz|vDf*x|Y#lQgpFt-t@_ZqZRp`nfX^t>zs(-*%CGOMF0m@dtXs{|6SFD8P~DH>36K
zOe^iXyTCtcJ+!z^Vhe7U!R;7RHt$6-cV4lB+v~Lzd$lI9BQN+4Y2s|CmY9O4#x8+B
zzB|bJa1B<-zm*iV^|(3S2BbGIM(}P+6GqPo0*7^d)T_(7LTk$o8c?jmrpw!rZI_>*
zMC=!=tE<N2Q{Ur(U4e{-%ERneDF`v^F*fuLdYl(Px2gu|8?}brdq&vnQaRK<`Gr&5
z@=n<IVHvI{a)OU4dL+b5i6smeFbk=A8uDlsym+g_G_tKxEcOz7H&I|34%#GrOh2bg
zPU6!yD`{i)39!5~k`+GEf%<2rtfo={9fAnE|9u0<?N~`9^uJ@JsWN{5CJu5(CKA!r
zNf;uO1G6Vbv3#;D$*Vg?R2K%J)6p+9rq`Ey=cNo<2I8!Bf-CyZlOS*Nc4GKwO_G?t
z0}O9!u=5sQ(P!i+B6qP4r`)_K@U{%Y65Hz}!Db9GPVW?mT%Sn{bG_)0Z3VRKZbsX9
zS0Znj1ACjb*)N{2qvI4t>{&3#bv?vF6aEaC5ecJaj>FV(vr#!^fGhY?!119A@Uc}T
z-DlJ=_YU8s;rFOM+I+A4@oDNibPT6`-;7(QhGUQ)&u7-)Js1jdw1CcGG9`-ad`=S4
zemN0dRV8wJqkx#!XK~7w6N#}=JiX#}fH>P~vYq><Fv{~-I+BjV=XN=E_G}L6`<jfo
zbUJ5K(8-1DYov;$@etCL3^51FxWJXEkatU*{U_8W$22?{EHYtv+U6|YwF2IsD#9#5
zJHB4KALMc#!2M_uX8Y9<ZL|{k9AFrHdo9MSXW4Sb3IwCeYG|N!IW0T-2Vd~q#sj;n
z!Aw;Li;W)9(qR{TsdpN$y@>$5?I%(BaW>7pqD$+}ZQ-WL+hE^@o%kej5`%LqSY_lJ
znEYisX#bRD-BqS!A>T`TU#Z4?@92Qridrr@P@B1o*5Rg|+yV;nzA!R50iL!<u#IZt
zSd5Gd%{Wvc)JrL*K?{6^<`JgYZM+epW<<aY?<;5m9dIKq6cm5?kRb0sPVZQokbUDh
zIS$uwAbb(6k~zw)oBjY!3a+E0W+;kn|BDh|chjc0cU0?emoVqm4v5#0BcI=zGyHAH
zvPT+_##?HjT$_&vD){{SxN<E1@f5$JEsM_dX7`&A!XK&<Yx7OC*@wgUuTQ9h^acK%
zumk>z?FI3XT{z#a3KT9~;9}*>$UC3?=$m1Ka>4!Jq0mqL^QW_SR{E&gzLrdh(<KpK
z2QjJd6xZ|M0pHUu$8p<Nu{W8TY+#IzAn<h-Xe>>o8_hRE_P0Y&WTMRsR~JH~KSSkS
zd+-Q(0V!Xl_z0j2d@eU8y5(Z{jqmDs)lC&9Z#JaE8{*);f;`I&b_IpTay$^qbEj`u
z;m{@xl9=!pC5@tCd(CAo;iV~Sa=nS6SL-;{8BUP!<{3n~y@JrL91PByi^h&W(Cxb#
zvlk!BEaWO+sBI>ze7PKTVh@l7*>UK<DvwrGm7`QvGHm^54flH+VOeb{8dTo_E%D33
zAq#b;quzp^(n)Z&G6WC2E5qn99w@3P!?ObtF!#_~VXtsG>GqX?y*&G><=ryyd7BON
zONQXWaV^r*cLn9G<XM48H|>68O|BJ8CBYB$nd@3R_zfSqiDnbn-%)<t!Tu7E<#T;C
zdp4q9CC}cP5{#iWdSuy!nW(3$z$#y<3z~xrxm2FnGjzikf`_{3K>S%@{#|MG*z_K2
z2F5^dw*)L-dLP_dr{b65+nBmh2cmmQIgPP8%<j*467S3BA&nnl_H%c98hR5SJ}tlt
zlgpqpv=M%`0oyE@0S6Th5b^cH@NwrFGTKdx%bHn+mR=p4=iVsv=Cir#32w|}_i4B_
z<_9!OUFWvX>4Tt^yDE-3hvI}aN63126%;$ahbYv>f##G#jB&08y&Qf%bv3@iPK{A9
z%Mr|T@*ir~*8w77OEJCKhJC7@!A_4qLELuo^N*r7D&ft~ZA{8Q(l`YA_Q?>nW2Lyv
z4fq{RJ#MlK#XV`M=<-gT)4pO${7+0{(${Lx>To#7tU4*IJ^;khVGpX#)*=g2Pm$_K
zcXp`Rn#D`zq2`b__R3ikgW}0(MMp4e`SZkQ>Mn96w4bxsj`*ou1ngyxp{;f=6gXW#
zPp47rnDtCh4fuk4)iz>=oVC!Wqnpd|8H6<NpPcYR0?1Eq#EukMVqU7p)@~fj6r()3
z{sW`ftHv}GUi}XyPQDBlSC@cKi7V8b)PXgrs<^FIfSpsLNk`TiF0$$wh@bj|dKz<B
zG$+m)R_%qtX@4==@H8keE67QR1BDh{OuT&qQ_JPCSnLk(JUW9<FDzj4-Ip<U*oBB(
z_;X`;R52>dlwvQ(C1T5>PuL&zLJ&8W?-JZELb39*<YG-Q5!-%-i?vh5a7Al+J8Kk%
zPrZo}g^l#?vL9$*oeP6)GQ^?$D#%}*N8IhYaN*23?53(Wv&|m^uJc0a<@OdBUJyr3
zzGah*3nGYFD!(s2Ye(ZP4Vm*c7qZ-A3RypQE-`musO4D<Kb0nNsgAvLe8Mjb+7nrE
zd-8HJAyb2;F1SY>L#4>aA{BTxm4avfUx-^`Po5TDA~Bank#SYQX!VR|HJPe0O{49c
zZc;e9?Q=tsxZP;AdKSr9@Q<F@qDf|E%wvVmO=wJQCU~}vfTLfFvE`{Ax6$1NtlxPN
z2vQ)?9XF^;^3T~Bw(0`cGG#ou{Srj)F(S1AJ|xWZC${d8;Lbce$xZOLV2{N4ok1vJ
zN*?}XS0{n^w+qO5^Er4&LY?Ju+Dv?vBnwWfq82YFq5FvCsA?Zh>XNfDW_b&)ZqlIA
zJ_q32^=vfTV9Gu=8nK=kt|Yo820#9gA@}J<ObO?|nOn}GobDo+es%)h@$yBHWrEo{
zmE&-2(j$m}(u(`+7GU+zDlTLFd)R9H8q2L7aOzq+c^_XZrx0`jtwssq_Uf^0>MvQ+
z9^H+p(>>wA3ku1au3Xn}Ew-$zf`uQRLG{@v65_1SO2>*Z%_|}>T<OofS@{<E9RxY<
zG?B$VIY~XV$C9xd%-GEnBFx}hJ9e7X!yySRvSHSKBGN3+$<!v}M4mUiI7^(kpO9yd
z3b(=VnjchLM2=Y}e?<igAKt-T2j`My$)=d+G<=sf4a`cTa>8IN-<3uSB&3<G#bh*`
zFb=MG@Y$7j;>_Q^1Cm9~(ZtQmaHb*8{GYiCaw6JLhd-mv{jH8)uAYO3a<f@@zz39k
z+6RlzTd>yNJ`{P}D6sYE#>!WPR1nmL+2w(x`0FBCI!l-BU9X7V7a}+=*aOY|EAb_9
zW%l>^tmOj(@+ViE#vTnpZ}DTO?IU0Y<43U7k}7nUb})Qv`+=ninW!QvOICWXK(i<h
zawyRdB-HQG)kDtMJM<c+boIfu4nBLc`3CoU=s7a4Nl^Y*gRDHrKaZK~a4X+E_;y!;
zyk9YaZG5>I-TlP`jd5~pVd8Sg0PBiG(M<flPK6Xdw1I?$YQ(2$Gg<U{0rlxW#3dh1
z!O#AAJgdBeKINSxxucC}<<MrKNV08tcZ??6wZD~13Jk@@QxfFJ_T@}x#Vj`MqZ8Y<
zgWrSjd-3Hg8-F!D!N`?sP`~FRs7$$ptJ5d4Z-*B^{c=n8@?ILmxl%f%W*N!I;Q3!s
zN`n9HdZ1JK0DmrchuXE1xMqi8EaurxcP=~AbE$k3@>3Yx{OiaK%-DfP4@W?J*dOdF
z8N|08W{|%?nZ@$=>u@rO_ivoSBjzvQrB){Oeo_jaI&ZjIt^y4<8sT8ma;9l6i$3w=
zSl3!r=D9<WxZe(k;6JOd*5w(}M-6<|TNAUENVCCnui&b<Cgb@}ETb|NJx5oe-K>*j
z{?`&>)?ZKQhI}0N_C6S|wV<o_$q|DKEs$7og(R+=O*Tm=U`fR-F26;LWUnnmtKKiL
zmY?B2`njHF7Oln@e^I8*Ovr&lx#+a#G_Di@A}y5yJBRq4Nb}0sos!XHdTRhQ1^AJQ
zXMC?DeKQwqcoHVdw?mfac$U&LNSA($#7PU|F#F7EwCm6!8ERX&bA>8wZo>u)TD+$s
zCPN+E&+q2OJ9hGaqX}qTUd>gO=0Vo%)5Q3w%MC-hJT!K0qfO<PK)uP4yf7KT4q3}$
z%wtIm+G~hOYSIul;2_A~wur3`J!Lg9sECU?!_SJIicuHt1kWuwjT=+`LiX{$@X1yR
zOf=u1y=4I$3{q#wJ^VL)cNEoJ7DLV&21AC=LqR~P9dyQ@CSKxc0<l^~J!bLFgR&;B
za(xeK&o^MXosY2mf(h8!DDxilO*oEc#K<RagQoF>{F`QqhBf(c-sS~9Sf79!iyMJ9
zG~(e?bD%iXi)ZW&^WVZPSX7pccQrlPfcAXOLT)@cdNvFkcbJpPgIXxFJq|i{Uc~Pg
zQ!%Q|7-wD9WO_QhvswN$$M?@z^qqq!qnQUCePeOo-F>ut#ss3sXA)K)e1>oK)WRS6
zKy-2H5k7aWfS^@1w0L8T;Pl))$k;KN*+<SJU$319-_@2Rw~A-u?owi(R!t%ya;Na_
z-EpLHv?D$^W6L5hih<dykxX^d20W-WfvbHzopT8ar0<>-qhEPG`fijUi+M)En+-nj
z(B~oOcPN2K!b5b^9z)dYE<*D=e<J8nW@0+i$?=#-o@e|St!fe=)u9YmJo%1&4!5vB
z*b9AZJjlD}0=n5pK$eb-L85L>M0&13_S9$aZv1V0<1xg2cp-;{3#YK6_37lqnGD!^
zC6;bqnnQ+1X`$)Vvs}0Sd@Sbka|<m4p|7C}n+^k!Tjc_W4!T0noy#0cHz5JRd{3co
z78(X#u-a)<fWr&KF>1#Oc%8ZjZ{MjvnHj3YdH-~F%cK}>jy=a8cgn%CKAaOdNyx^8
zwK#Ug6|7s_4vv3Ek};nKIjJIV)YrVuMN3h5+p11BvDxg|o?5JTP=Y-M@8RPcc^3F%
z56u4=ioNSrQU{xhq$74edN)*p=f!RK+Qbd_$}R`XO{Ux{Q9$9RDPS|Q7;94!G4*;;
z#qk~^=5*>CYFHLx!tZ~ykwsB4g~ilY*N<kppC_xg@aOLBa-fiO8U|uSm{Q1c((hdj
zUk9aG1K+Q;SjoG>r{vLCZ*8(y&WF!($AMebC-m`GrUmbp&_fsQkW}1GEklpO&sAyy
z<>33!l=~VBitSNL!b~{MU@H14UWK38K>Wv-3$r9dxsEn>to;0fwtqc<3sY+`JoYl&
zH{kmYUe;uT>OFjYaDa1ZSxEIArm^6YB@j?x4#_==Y(n2x)KBooMa$dJD#ik51h>Kn
z-eHvC*egh|u_GqRTksUQgi@Jjsm>Hf=votoVGgM{JYzQm{gh!peWsFd&v;sQ%81oX
zY~+qmGh$P`kSRzkhvmw(cwJEq)p>7ATqxu2XY|plZ~)+s1@TdkBRkcAxd`Wg>}d;u
z|L{p_dG-vs(cw?3b5`JSmml!cz?gX!HK5UlMwqLyiCL=cCll<Hv3tEE^Ee~UmClgG
zI|o!5)P%y<J8AfItOb**s|D#DRp3!3g+t!6*|$&n=(oWbXI!%**K0$t^|Bau)Q#e5
zzLT-(^8}Xad=~2?j8SX;Dsucp9&FwqM{E}OL9(?3F`TVK+NNuf=Powr?3j<zBXZFC
ziaBwdIRm@?6rf*qAsXAyK>>dc`u!jR`+vQ`y~n<zb@QNbbN5fWH>VgpEC(^}!(xzX
zIYZnv#9+(SS@`smGV`E37pYl>v4>L_f6yn<O82=7VHU)GXc=tKmLf0i{o+K#iLlq$
zpV+3}#=8Gvkt$b%eZ37yy41wG!NOos(OEM5FBa>?@`aJIE5P&HadId=oc73xuwqeP
zdeLYc>n{q2cPqr%P2PJlTqFXKi-p+X7=@a*q{(Uqq}ewa2pY0tnUw-M`5x4@UW5W4
z{vKJO&8!1O=*WUAq?2c@r9QHuZvGbVh|h!T9KXl=xgO9aw|dysz8FR190g;h%7dSR
z8noV%V4u3e;K@&2ymw9)|At7D6|w8c*3bf+erzO*OCF@s#mBj?L5{F_MGR+j_zyLk
zOKEOqJ4$AcBLD3+XGhA*xWs|cWTQ$OJtCkq<EK86Ue?SN$4_R;&Xd@d`MJ=&uLX=G
z{b|No87|pcfr{0vCFQG?8MopXk$&NW#f?+hyQnlg=vsnCck0k>QatK!wqwt0`1|d)
z4lX9%hK0;kz}~?bT!5_zI~J^kolURs(zhnmDV@j+OSDLLsR9!ZIEyQn-=nujE3l?W
zo;7u-7o{Y424MJTzB3UkoX{cy#pASDri?sZKKG95M~p-H<xybcKash{Wup1>*IcVf
zFXtsyhDV-#!@oHfIk~%2aPp6<=q5Xb=kr_>JkIHdU8c5V?9p<Fb79~bbson(7h@K0
zYM@no4*BvxlXTu5NlkMu;48v&0rbYu+m9Ntpe+qgnH!TeK0G7${c@7KMjgv0Y{L-m
z+g!`|aFkyD7F6yf!@SWHMPy5FsN0O=LX#zk!+u%%ZzJ!5|K5k(G0H7%JHU<GCx_qc
zWLegjX?S+n1GM=b_(b<o>`^V?G9;zh%ET<e-$`an+Etyk8t*2f4Ck|n=@<A8_a{vM
zxQOjPlYsf5d+EA=$1$z15p8EFLb2l`JTPZ0Q90a&F+0XHeNZRYSOqM)BSKQMJRooO
zZuHCdrKyvGNRvhdJvaOo$G@$Dj@OlVI_w*s*_Q@0i=3%$`%U!kyeU}icAxgw3z+n@
zB!Sh6K}>6{0%iD)$!ex#o8lId+GIy+^+fSbXAWj`@s5P(IFcGSl39K`$#rik0hN|V
zIORzSSVulasm|}{`IP4^O-aMek~g7hy(JsHtq20R$z0igWf;E6f@q!E2le3}1zoY$
zOndTM<bJdRQ5r$cDg|L#|6+W8?mrNCUPP^RCb-ey42F5~?|{~|tgKEG_w*W}Q^#8_
zLs}I~7Me1TC@BP4ZKisO?|yb&$KvZv0_Q$OcEX|;D-Z6&-28Dk1>f+O`{nH2W)~LG
zsKuIjW~Im1Rm9zg?=OyihmT5*p`U0Z-EKaXaOTIL<NYqq;m|N%^vlA$vJ{xc^BGo1
znvzuUbcmiBL`0UQ6PumKcq-!%K6M$--1L8PG1|BIz4sFoi#Y|xiNA#P^Sy|5ScsLv
zmoyrGdVo60#8dZ1S1x~`9;J)Q1p_C}f#LZ%#QW+A943iGvL_y)T#GCYyNZD?#kjsS
z6((}>9{Svuhd=(|BrQvV#2tHv+VMw8_rizJd&U)xm!zRWg9HY8lnOQ-F=SqIhGB0~
z3OOJfiXwiwXuf<rev7okC9B2}!KtfU{f}0venA`Zc^}2k-2HfTMH`pF?V-I=ZtUnm
zU3Bf9&$Lg>Le-!ATa@4L$+sCZpD-r~SK`?13r8UKcMe4Hb9BYQ7ZuJYpF(P54)%_J
zPNU>!v1Bv;H?6J#=3CXU@OmoGzjh{}zeloC-$_hvQ5`zg7+~DuOG4F|eQ-;%93PE6
z3yF%qIQ>7mq{46)h|bwV{v^cW_zSx5G%1j)xYB~JOR`|oX5RS}Cq}$=t?^@b0eS`K
zF*n|KpSeXI^EV&iR#=78Gx_}f&h9<;?}`>vmN8{#uP<j}z24~ZydUZc6rgjh1Zz%K
zU_-41u+-NC{YKuxWIBS@z8=GrEmG-bGbyTjiRW9=3p6C-3~oO^3gYta5s`iOE4)Ug
zk{t@8$%mA;oMe$ZT$uS#czqr3uhtD9-!6v0?oTpMf4K_`vz3WnngvKKj)JFg+sLhn
z62v-N7A&4#0BWSb9tJ+(EN&Zs*rSu^uCoKLm^^}KwGCV}xew>hmSIy-I+XS46aNz4
zqoZdlykI7Qm0XqJ`58iX$&A6Uq+tBDaTEy}t4Kypl_vGWSr{$yk0y4!#vP6PS*U0M
zZWEDU71ssCaQ7;rc))|$t=j~8yGtq-cTK{*C2{y^&U%({bcEn&vJPpoNave>w=k$<
zq}2d_w>`etl%@2G5%)qBQe}{aw#O%;+0uVpmUa#{oQa1RM|D<}Z%I6|J8@ZWBrLg?
zMbfW+rQNk#$ZMzXpnOh&Shw@LV!rEEP_&(kxm!R3+wHjpJExP6!<$*1LKAvESWBFD
z$g$UE23+V#A;@paK=-9((5SwZbZd_wU9ZMMR_`-BP<jSyS4EH!?@z*-wNB*oHfv_&
z@&eMIDznDm6jX8W#xs_6*xb7UzIBNa|0Tgd6{hlC8E3Zg-%|4J{1eViFApw`e}m_f
zUvM*G?{f<#8L?pV73h#wF5GVz1Yed~<Lf~avRz^u^i7FE-{$w|>|nt}^0(ixxfI0(
zygE$Zml(65FcUVheLj(LlO*0r`mCXH9eESGh>g8g4+bvHpuIr_C&h%5AO-$eGqY!h
z!oQ)(p1UBkCk7M+PXwLJjda%Rhk{}^_O{59?UfW`?w@(@a1igH)R;`_mOFyB{ab#9
z-^~@OoTKiRyw8t+Kg(^AA&H7Te2+Sc*7~0!$(a`H;&w^aefK;hURcO_((G{8&o_8M
zrVdJP@txoU&jok*PHvalWNdc0i1(iJyU1i2#_9gV!@B2imEUWK-6TQ&>IUJTx@pY8
zWHx3@{!D{CjZxKk2K)3O1#%<53aV!Co|AGpYF!$|sR?38)4L-yXo{@Bqe~O>Ce2|V
zmUq#bLRU6w6yNc)=*OBgS2TK~Ks4lyuxlS-oxX|O$y}b(QP6?WOFvV$yYun#rfibU
zGmSlp|HHb|U+Lzf>E!wd{vI{ah|H6oP7Eha28H=S`0=JYY1^j9;xl96@ChYWx3d}_
z_guqn_M560$ibBpH|UD9#>`FfKO8sgPaBkep|M7-FhDhu$c-OO7MF^#4LREQzGo?3
zls<tCs<&~yZ3cdOIf8U%=rN_8XUM+;P9*E@J6s<fi9aGIk$`?#sJm1Ty_^0D+%Kep
z)IbD2zFfymyC_PUMQ0N6x3@s5WDg$L$UB$Hj>GpCSD-p>i03Iy#@}y_<NW<5q;l;A
zs1u9B&f2Fm%XK7jIs5R38NpzyQJjix7PX$*D(oCqrdK!2fyoQPAa6$zz8krQXzuri
z5oy&R(rG_C^Sv4Ghq*-eJWIe$f99dxq!L`<@`@^pZX+V6v`IkIO*|336Q@rcL5@tX
z;a;g(V&vLlaGc~w3f5-8&%1GawnhaEd7jg*zeyw`Fq0^HR|_h03G-D9B0ZIQ;H<uv
zL{*)p+3Hp36eUV7$;+|rgG)#_KMzb?SOL<%TDhaQPQ$wQ{29IbG^VY(h(4xYAw;)_
zmZZ%f@AML((zCf@T=aJQ`(zk@yWEFyVZC&rP6yJ?5L8+lPVDVwvqzsbc)#3S)H@hJ
zQ~F);_;wj~tk{Phn$|~snig<}lL|oF^#~-MX+VpDNDSq3@Ea<IP-hL#Vfiahl*dLu
z=oNSJH#ZWF7oDT(y3RDpT8v%4){8%uX_3zAAl$Hi9ZbB0I3{ugRSUGj>w!ra_Us~k
z@NEovx7ZNg!wC@CakpZTj0JrdIEKYuslop8Ub?@#5VxfEVa;4UrYw06Wwf-&)K?w6
z`zjC@9uQ|DivE_?X<58;#fB_7Wq<=GCAr3MC8FdvlkDPsM-HEO?yZIg-?<#ZmxKDG
z@3j*7`IXNAi`e2ot1T2;^6aXa#(YjhlHNU0ih~9k%*x0TX8x-|x#)95wmb#L1Pjrq
zD<6aT9B@#$D9gLsflc$C2<=W!VHJ&A(R6DjYKyC|+DHXl+apTe>YXNui$YP}T!sbi
zm?4<tnvDbh#ev1)3;3V&60-WeJd<*o&g8-+QM$E3sJOm|S}zL0`eX+tc9`E8L^`k)
zA_?%WZ!%Gp0DQ9Y9X2E{1MQk!7^<hkikcEQ*-;^I)7lrkp9xXJw1jHA+M~~8ZSpU8
zG#mL$3|^*>B~pu=ncnz2!o!tgneX==DiRn%U)HQ)`_|dBqseozcFqX2e3SvV_a?(O
zc^CX7Y=-c*b9iLxI5=&63x|6DVz$Rl)~_W`vL=Zz6A@FEIO{z~I%~7)BR#ZUQ<@f@
zo)3p>Y)H@<V5a{)M7H8S3{SWRSu#s7S>X!YPq2h>53KOad=AxkzSiTp8tiA&Iudki
zRfQ6|liV<UawmQf8ufnSEWZ9n6_#8e=Ns*qyp1aPI4zcXFWUx_9()JmbUWIhnn@mN
zxj~&(BvkMCjhmwD!Dvq`4#g|8+w%(1bVd+f4Y^LXEA{Z+gUQU+@+s#Dk7>ravzYVz
zGuBrXW6<Fep}uSn2r7!`<A$er;I|q%Y<dGr<l`|@mXOwCp2D!x2CR0+cl^BTJ9PZ|
z0L6o6xTBd;5YTcTyysWKE(;N!CoIYI4c_6-`;t&M=DsjAT#5C%Zlhka%(>mKR-*5%
zbr=}lB@EwdP2;9~txz9jO0<r5R{T0B%F0aVz=VZr?D&#DXj^^&`F1JuU!N#QNIr}n
zXYRt!oDB@@<}=?b8L)ZID?!ia1?>2{`)HC^fY(k(6McIXW`3XNaZH#@GRjVpo)uBt
z3=dn@Y9o$bkCSn-*?i`b+r*9d_8IaDN3c6pJfHRFX88H^8f4{8B%>tU&^ss^x5*o_
zP=g>G;WimXoW_&Dt3F&oZ7?h!&|sbIKe#&rLuPnxCrX#A(Y<TMFjRjK8>|X2BC8)W
zWW|KuH-6&mqI)oHaXCy#v7ySHG7vPHR9KIi$^Ca>BkzIF#qx@8RDRP3u6>RV`MXX)
zB2&EZZbLJUy_|xUJ7<B>*j$Jdui>Ir6@cdDtK3q#IqdNBeYkT$ELJVLjpuWnXj#M~
zd{iO8b#AZl`MeMk?sSlrcF2>0P7`9f%?j&ohfv#2WeE8*L|^(`!Js?2Xqm6ixCs9H
zf9(&<nkU1O_c+4b5o5qM(G*7;C_>AUu{a#)gLkW&uzN%mUB8BBHP0K%G?%wSz_PCx
znk30QXI;YLz6NOj5Q$x7qcP<3Eo>V91NSDq<+SZ$LGFMe^A8mA`@>me$Dzlx<GMAv
zh1Jk1xgEqdF%~17)R@QySF7cm0_ks#r)ELx;8X!29f$g1_Vz+Z_`<WkL&~_}#1cBl
z<Y7)3?^APG#{8pZGICg)v=}-NK7@q5_wQE}5B7onEeG~+tR<7rkH`6|?qQXF8C~hj
z-wpceq0sLvI)zt(V!kJAcm9BX)0?p~`7eKF-Nf2$C&DfhKA-d%as2Nt*wb+n`xK9(
zSm8v>nD&|W-%Ll!?*bgP7eGs)0}QyF1%=6B7@s<kRe7BtwodZc?Gr>YtG7edl5Z&S
zdIz1r_fLC!RnYVMJl39ej6~e-#LFXqWlG<m|5J44(NunI7)D4kB_UL1GGz)m``Jk;
zqzp-tq>?5{lavM_BqULYB9c;)B02A~laQp6=6;fBPLfm-^}XN!)?ztly?ejUbKlo>
z4t}8OgFATZqYBl2I*6qk+_BlL2t0OP7uGFaPR(5o;6$gR5V1*-MK>@qy>u##<uY-z
zGm;FL3NprN!n1E5z>ix?=|gZsiCY(NNKczpc0VTV&%$Vo<!zX{{3L`a%46*M%lH~w
zxVnAosmV%NYQ9`nux&*)MvYWv556sCO?m2c)NmqX?a`w5K$j%Q@R{{DLNwOi#@@`2
z<ML8h+efdBr~7Yx<&yra28HkjV%;>2bd)~9_1`Ds4o^?yP3q_{?JTX#jV8Zb#hGy9
zXx=Y$9X0d^aB@=z+!(f?uBHMi%b7EqdJ9^i!sj?7ui+i-xkRCKAO9TuMEZlHP=&K+
zPcpqxGitQp?X92OPtKYv`8ttuuRHOC=TSVc{|JgUEu}L2UShqT5SCrPhA+PHGl|U;
zSXkW$xYBF_b^YIA;Vhn?E?PvVUDl!2OJ-ws-Z^Z$J&CQpP{{9o?r_1m^7PxBWi+>Y
z0*bs2A-^u1h57TUQ0KV`vsB2#Vc$!f%-ZSH`}tV7J5qv)eme-RlgwCk?r(gsA`cGy
zkYE$7`1#399X5;KKQ{DC$E&t^=oP;mJuYPks~*Nvldc=6Zp`;^59m|H>|SCz=Ojw|
zO~&nNax}=X0b@O4x$09<=;$`YKlcx#$?X5Ib@VIV_wB_J?ZwzSTMIVMXdE-w3*g<`
zBRRJv=jheR*C4U09tTr7R6Z0)ZG#aaxL)CI!&6Y7+zvhW)#&|0$FY2l41M&Wno~N`
z2>b3|!uOS1h||Yf)b~3ChpiRqv+MqJ^uxvI7kmWt7V1zgSD9<sm@16;lZ?S#j&$py
ziLCd~a$@s@gDc&~adnz4e5gWFw$78YlkMa=ttxbtUMza8n?+AZNz$|(8Ti=Tn|Zze
z27a1Tu-vc(&fok?R72YZ))(i&%j+s=*(}TU_20p;m-4u|(UcK?QMRda5-rdFF8n2?
z$Zn}U;vGQ;P-EO4Si(Db)<pR;o4$7>r^g<Ti_C-KdC&O_a6HOvaiLb_9k^q`X=sre
zk86z2aG!M~X!PQ*uyl7Zef+!$QPhTsy`2nFKWwqC>=raO)^i2ZuVAi!4hHEu(ruB>
zG{*im6c?y5HNQk`+5A^9D_)81%-KyzgA;nH)nN9hTDZQ8B8mF}vNFbOI?oT?nc@y%
z*}$b~t!7W-U!(MKdzv>c5R4A-4#O4g+(wgp44a=tv-w<Luz5AyUhWV3ca4SiUCQ+2
zr%`Mq-~YO*@eOz0UxtEtvh1sv4wxtJ!s-KM@FDk)(6@Ia)Be_g+3#x6H_8rDwl9Lr
zIc~Ia-Ug7m-vz!hP280miD36P99|5yfc2t82(tSDn*|%-UW`156&b;sVoeeoG@6lJ
z7eTtBibQ-BWxTYHy>WB_>%neNUBo+IwmXtqqsyqeK9uREMAHKm`!M=THt{*~0Gp$D
zZlB8+9Oxc|B~N(93Gd-)yzjy3FX6LyuO!J}%?I8iavIBD9LE})3jFT56;q8n@a%`1
z5GY$oib*X#_~OCB^7e9pKeGfom)VkaJ&!oKQBLqIbteS09mBQ<@8Cv$KZ>I^t$bWf
zu8wtO-_vT)e6unQd$5L?IJHCc3vIfLcQ3k4aN&8YBKUq}J=fJopl;P^uG@4jv~5~N
zl?)$a(XbO0;W2MLQ`701ocBVR4KjoboW&yb#~Ae0i`5)?gj7wHop17F(iyK{R<0;Z
z5M4?;2Hs(lQzVR`LYSeEj=suc;Pw7eXr7-&qCM|$_1}3ugu9TaW}4Wmye~)d@Jjr)
z?lL}GbV;aTr%S&iCSk_FC>DGB9C>IfM?2#+aFqLRGP}14C9R+0UxiTGx%dzl^7bU2
z+P#Yod?3t?zsu|O%dpw}tf;;4J+h={pm$CJo_0|Tiiu*DGwnc9d;^^}@fW|xkEGwD
zCeU#y=5%n?RE)Yu$Of&W7|C;U+Y3vuh>fMb*DP>db~(PYvY}D8@~d6GS>dErMmW$Q
z&z)HG6<k}ZsQI>ZVOY`$)HD5u<q=xU?Y=JaRM%#IhY!)ShVtyx{YkiWQZyQbNiyEA
z3471$vhkz(;aJsvjGY{bL58=<dCe$nUcCe?i70wAU2@d!4Oi?i8E2&LhLC%SRMl7w
zH=no!1JfA{H;m_;A6~~kslTWhI#N)6bP`Lu@`qT=l80VxXHxZM12={D|BcTb3AP1%
z|F%(`^@fyT=4_tP^mQ(54w}m<X7K&E8S5x(HKvk5ouC+ffy}C%E}T`Y27B&g&?7;?
z7+^08ZRN4V{AdX0zp4Z;To0t;!<MAZ;y=t%P-c>o`@pVbJsi+9Vh3%EFzi0>7rt|!
zTh`$KH?vlu$WZq+BkxO)G43Uq5&jfSegxoObz`*G4#IdS0};(%By~Q|JdA6EuH6Eh
z=2ilIA`x&}%?j#tQ?b%lhKAla%>_w_vv*n-anMwTB}q^gvR#s$&bY)yZG2<jFhNXs
zYJCrMIge&f*Dr(eh$|?5SRLG}($HhmSK_a^nfUVF=rM6mQFus>y1o`;JGn_9%^t$N
z#1$~_pFVXFmgBS)T2x8768~J$pnHvY-)h)+{+(eRhR)eU{f$C|Ym5^F_U?7CLU9M3
z6!jeLRq)K_NjL0cb8Jbg)&mSU{D`Cq=Hr#X*^o4Bi;bVpQJt|{>G_#~Se0Qe-1}k-
zOFt6<tvSzOm&6F-v9=0~^cQ1eO$1i2<7Xsgk2z`041g>z4AeDYKJRVWkZmox4j-js
z>*blE*cVuJ=rFX#{R0!5dl>M#g-d8G!sPj3q+jt3w}?Mu%=FffZ%fyp$+nFG2c9W;
zs`nkYRCfl>J$(`m-G7Av|E)%YkGs$!s};81yU3|27va5Ihslh4YHWG?Xbc{ElV@j`
zqnqY?Y>YZqT^dpf4PFyiv%0_VaH&2Wx^R`~&N?8RRJs<;wO`uzPQ8e~H}L+eUv}XA
z;~*r<k7W0{9z)dR>r}%n8+{H&l7$+)FZQgjAbm?0&+q*Qx7Ws_p<*f)>=S2CTzXO1
zrOZbE<nOtauTU{omW|n)j^|Y^*`>X4@b`5bryD$uHvQUyN3EmKx4wh3N}tSBZ!d%M
zIt#G;kvm&fvXQGjQ;yH|3?TjJZcZ<L6k9*umX4l$5u2O-2|k{GfDOm)<Irbw^!o{H
zuw^RAS6W4#H#(uxI7@-1dndj*?8AZz0%+_>4Xo(S#MeCuobPsB=9e>*ofq%KSL#Z%
z>ESV&wp5HkOczw1yTAp%EP-8y`_Or59d~cuMvUdN*r}&K;;Vpbcy8h`Z1Iif`&DZ=
zKa2Y~{<0{T@3|oC^>`@|KP}B_;@=B5^3O3@aY=R*R?`tQ9RlqvIK9s8*c7@M7Tuph
z_g&qIcJcB=t=WlH`8C7!H7`*yJ)eJXS3}eGN8A#v$+S;#9DAlhxt{%5u)y*G{4BL(
z)r+>$n~LMvcHMEz+p-U@MHb;RyRBfbO_5G&8%@g#)cM@*Hk=(5!#h&MA(Hpp<pieS
z{A_u4q<aGU*>VT|?9^r?xCZI9L9oB!33Y3q;I!Kd;J}GfxMo7Ba9{0OsO{524Q1Z@
z^UsssyeNl%!-DAxNfUg@_rcy>$cL%bJTE|KK>gmRv+A9O7?gYr=8_k%*Fgq%k4OXg
zn-=UpM?2hG7mr5*fE9-+GdX5LA0_=JsZ%})A2l>`!ZH_(oOq9STk-!B`!A4`BF()z
z_yuhQ^T~<ZYcY^#jQMFbU`w$rS*xzX3Mw~(+%5@rCGsoxtnn>G-d@fgKevL18#Ne@
z?qm`3_7ah+SzLFA8n&h#p$BS4u#i<>aMa~7aNO=W_d?8=woR2LqWy|E?Y{=N8`ww&
z2UgQmwKmK@B}NCN=dzXj{3j{(AmmJ4M<0EhLJU(>*}j$P?CE))JKZvZ9h{qpIW?P!
zs`?SGD!hz)(I86w?nh(2#czDJtP*aiS!3Zn2OK*;934EA1a)R&{F$GIsupKByIonh
zB=8ZZ+1!T3)(^RDkJYJ__8UCE^EwnK%!9D=hLE78A>7@^yYAi=Vc0`=TCt9w1vNYY
z-CKM=<);n=KhYy(=MeXichoFfW=OXWHVK=HUSjE$GGd@R2rI?Z>9(@f*yeK(=YEW*
zd9%f+G$!CI``Pppe;Iq3l#GubOVOI6V?a{qfL5vFDHU7ID#V*`DDNugzpn>`J-*mq
zT>~>5Wm)uE8EEyX#8!6&(&Sb`d#a~%-GBU;S!5e$|5k&od9H@B6(USze$Tb4J_q`X
zzwatcH$v`aG>zV3#gg<F(Zc<aSR}dv-+fHrp7pQfp3HR<JZX<W8}C$Hd^Uy_j5K7+
zX9i$gAS32_x@fZG2BzL!O>h4A3%vv4TtgAVPesdV*w=7Uj{IDHp&@u%|3`W?q?l!a
z6CCElz}hn}z=+mc5XE!wCQQhKHU$B#T#!f{eIlr8lomMF?nUWV9~$-SZ*`7xE_}26
zjLy+FIe+;)uEG63&eY0)?delxM^~reGJ^xu#9EY@lj|fPMUKfO9l(zf?rcSu9$m5R
zDVjbDq)wMa`3ZD4<|*9Al$9?r?@o;1pPLU8He_IU&_@hS=GkcG?}Vu@>+Sms2C;|F
z*X3d+4lFp1@1`VS;5k*ck)I<B+<GSTDp8{zq7&#~HOGbJk3e;!HnP&<B21j~032tX
z;6CRr<(Ye@$;Z2$VBvHEGH!X$6)F9Yc<>nY`4Wveja%?x#&q`Gfp^%1-b1S+@30~A
z7#-Yjk-XhKjm>xvhH?C}Jn;1i;lr<|vFXt<=vpwD^hV0Tt^qBs<MDdo1dpGveXbFX
z7+s24z8+Nj+;f-?SD{0AjHYf?LhY{WSpVTBP7iyD9;aNn{x#9i`|cQ6^@T!I)fkee
zKbPAxga11xOy)M`D$(jyYG7+Mk?G%CN^=u+@Kq4cWuL52Z6=rmqBk{}!=KOBPImHM
zEf-7X;eAJFkwK`cQ7UTcBnzCb--BzqkKjYS0HdN@1=AYsk@NV<J>BY!z8l_yvAivf
zni|0AFkOD1y9;;8xY3Tw*M$06a`4IhI&r%=i7nx@nQQxAcocdQiyWj$pwL@j7x#c`
zJMIDlt8WR+75(uwi^ZHPS73a*G+UWB16)U+!L271S=^?0FqvjR1$AQd-KYs9d0P*P
z@3w>G@DUPD9VV^8d?&Z3nd{&^WS)xg!r-SC=-@Py8F^Qr%!AMP!}9<tsLQcEgQo1^
zT`wqTxrUq1ucC^w0W{ENx$uMbXo1mQKkPer%>KwDYnGbP0A@ALbj^=9L~`&H&ysUv
z;c;g%y)F$?vo>I2w=Apv@Dj}nc}DOvIo5kskGy*BhD#f((L+~)1UY;m4nHPwRW=4}
z-7G1Z;rx#?=6Q8PD!O>Ek>~Or^hdqIChowsC@lKzg2f8J^)?UV9R3+*ws{w1O0J+q
zUD~u!Pm!ASmk`AXm7tW>jjCeejD=suw?_-GuI)aYc&i5gE(x$n+#llI%Q0$GHqCLd
zKssVEJ%2fgpYyC23@9HI1llQ6@y034TKP3MjX&?-iHoyqH=g1^ULoo6cc7(_A3%}k
z#add7VCnT=u(rVhH+N~V@R4V6f@K2jn6M0HJymB>8rk6S&K+{!$D;VkHqtO>4`+Yk
z1kdX^gNItOai^{qTfTf8to4uJ`$ky+Q+a=;-eGX?JOOT}NAmB*e=$UVJn0;8z`b41
z@q=1}fXK_Tkd@!bs*9sgqB#MV%r|E<*S>^W>kVL+HIKv%)InXJJC)2TBJ&?vqiEMl
z&QDL74X&DkualD~#SIvK=MjY4Zo;`~ycd1jRm|STaclJZk!*3|J4Cn8N~r*ihm4tn
zP(`qI)N44(pApXv1(2GrDJU%;j2CP=(Zp4Xl}{hR0*^l<&P!$4B2=ZPy2mnK)i@#^
z(??Pt<YVv~JEkyo0`ZM}EWG&bA~NMVOlVueWk!}`bhkO8+X1LKaFG*nbmXc>=knjx
zv25bUP<;ROBLo{pfh5@o-cz%=5amz!O<S9(7^|_+bxGh-@)|`t2krk1{o>a0%&7lj
z%kj5dJ2Xb-z!kR&QajraM!6-@LY)v)a7;sOE*A9*cH!#QIGP{!0nX+efY`|wNcEbL
z^k<AByBf@A0fQ+U%CtlImdCJ@pTSGj@;sdlv9RaF1a#^YCz=UrXx6z4CBuJ#=I&<h
zeP0dE_fUr3q|qd{cN_QSYdo3fkPaq9ndKO7rM|2dmD4@Ap(oO`I<OC&Jv}%B18w$b
z<#^J4^94pUW?^o34iu>eQE~q$986dM&nBs(r`1Lpg*ouw%?=>yDfFD9CacUbVe<Ta
z+;w9u$qASV8(c;*(~oL=_U<H7F-!dEx&_xRT!G8`b=ftCRWy1dFzKN&?6IF1HFrFV
z3;P+i9NvfL?k<7GZ7#e=x#8Mz@kUe{`2bQPGeEk%n6rEAga!w^AoQdiev^}6U(YVa
z@u4wP?z<)X;Uqy*SFgt~HFKtD)dQ!?jd*X08q;^$PJQQ$rfaX&VB9+kreQdh$!L1e
zSHA0Tf#EJ(M%N-YBMU#g%z&D;I`mPO21O-42R)tP1PeL5@@O<u$jA|VI&+!J&;1Q&
z_KUD-)fQ|QpSk<zZ$phfAI4%@C5-yR;jVYPV5riX&fe@sCb^Esa&i%sbfdA!b0fX(
z{t^cpyjTdI>nJ+DnEoBk1C6Uw*x|o*XqW8=W?Ct{2W<m1ULM6N=fohLg;=ei3f88^
zY;ebI_!c=5Q<9aT-+U_g{BWkev+Xft#8$fJ=QtWgcxFyS9k=ULF6WyaOGJ!5VY`zv
zSF7QPW^*;6(0?0oRyC)AIYPYt?lzibi=)Va`_<m2p49nCIiHP9rMv2*(E8#W<`=dX
z2wOtUrk9}oq?6=y<z5W?7>Oo{pHVV+5nfR|!{-DjUW%X3Qpc&%M>h-ce9<s>`Jf&B
z)Vz<MRhwdm*+a6=H4b<5{wCtKCN#fuA=7+iiq$ekz^epdgSG{`%+F6BWIRXRu5{ty
zXQgzE_a4|5e+ZX?3NDk9<@4U&)glg(+_ujm)ak-FDmUAmPOdRyB8SqdHIpN$h9ZZ3
zLQ(e0;3_=xFTwcAcbFf9H0h!d#*6TM;%lZLwtGD5pOr)dwQi%B%5GGdV#q94o`jc8
zUKl@o9|F@kE^~7%-l{o=W-3$I*(3|r|I(eNUpqwnR8GRs<apZtUIKhYk8y3mg!M16
zWc};<(W~AERsDLP*zj_7)ZGrw!zD+swJwWGdiol2UlI&*R%E_)=Rl_Y7(moM98>Wg
zKV3b6pC>63bw>eyw5%b|T825*lnzpKp$T?h@_=)XPI8V<Ea2Mf20TCFkDz$WFwY$o
zWwqnVLEgL@G^A!h;!!mgr`(7lk<LQ9Lr-wcf1+gT%~0BBB89c~iM&&!kyF}UiZwFL
zB=(y=cjB-&>i_+TCih~|`rRVZ5fDbAlH)*CUWW$QR*<<13(zQfJlW<i3aw4S+}^ii
z*yYnbBzR*W)V&%>hn7@v8oKf<b=_GSyj+CUth*!p^kW>XTIh*+f906?uqG?~T@0h1
zZlqyr(jjSO5>c<Z2p?iA?Q`6vSs@1E!;Mm`wNH%Bzh%k=zD>nMuk-j~|5GeZ&BpQO
zOEB>dpA5dnKy5Lhca1%80nb!-%f5jRI+n7%!B24Ra1g{zcfqP8Kq+u!1s=^<G*SWX
zhO4l8B`5YIV;Dm;FX3wqS$2QPQe3-N91}(MV{iEvOgr-me)9Vi|JaWxfA13nU6{$7
zl}15SBhQSIPoZc2z6ZO1I!rme3f8&K#FzE@?B`NZdZAH|E&H7XArVq6zj`-Djgn>s
zh9b;o<7;%+vLi+Ze`D{&dc2amimrJ!nLFQIE_4-~frJZx(0I*%_;da|h9kUDD&QS-
z1W7?<ycq^+*K$Qfj6T)Mg}eo}WXM>NnY_=%)s?a6`1UJjz4IFRq&k&344ATC?&)~b
zNs6j&vSANA_tL{Y1K^+_jXtKcsrhT6;IV8z?_LR@gLgF8_QGn87*&AX?*R5#FPlHV
zCX?F-J=q$AMX>(TC_3;b9TS5qQ6YO1?Tpr@i6(!s?)HApXi+QYaLSgdoiFBO?;XKI
zJ^PVQ8F8wWJMq(DcPf*d11GZAVSrL5YI!R&yF2v)ow%Kl9J3wWyS_qkj+1a+LOE`|
zJc@Twr%)dQPc~gBfUPsMXo-$2ZF1r}UrDZLd1fkoVOzkxusi`p)kciZGqOuT`>1KN
z8qHi8MZP?c0pha^s~q)6Q3V6#WFsp2!H9{dsM7HbPq{z^AW==jg8TM7JO9W{oGp5i
zx+(R*ES5<gShW*x(YLr4ra-OlQE*xI7vD;Xvv@sen)1?>KDlLw52aVbl?mabBV(H2
z?FvzB7d7M9fhLj~@<J#Q`=0!5_oPOSQfz1XXDrJQB_}4|<1|L5V~aN5D_v<zHjmm)
zMV9AX3l?2KoqTuW&Wb_2^z9ACR!*QRBY`dOJPv{L_mQ2SW^+ecHqyQyXUJyF!yxhT
zEBrQ;2947{F<vDBT^+M9oGnBsj)%Mp>f~q83^pp(m|Z%*3<u8a<Z3g6FgH<#HZEHU
z{;yr(kHl!IDG^P77X5(Q$Lr`=Q$;qI-ihtgUD)Usov>5$8P5Bpj*ZLXsAro3+q(2K
zgz9P1NY{EQI`bm*Yvo{GxGIDBauD_`7#$W4S6_eq4Vmm0j8nYDMF?)-w#TvH6K{<*
zvNkxh!;7waewbT+aUS!%EWtiGeZqA|DzKSn+B%HbOv9^o({#%RIJD26{!9cW5m5|F
zT~8Breph&V$Q!pytFl=-pU?q~8FyNrJCZVj?O{hCy!aT#6^gUIhrEA%T`MH+I}cbZ
z#_|>-ij4`PpE|R!dt3x}r@IGNuU^XT*rh{b%@pY7`CcPNG-92NFa5P%52OADaZ(bz
z!+6CA-bZH+3&)?t`HPn_>%u%ZRL=Lp_b9U3y+>*4wOG=vx}UbChj4lJe+8{hOgMC~
ziwj+v4OQ-AK&&f(s!NZ+hh<S1_fM1MrCuS$Md!Jwkc(LDYzuX7ZCJa44*nXFrV{r)
z;_VB(lesPwJ3n3G5~EdU@vMWGdgdE;Okc>gtzJsoGmmlsqXszt(-OGl$$$9wVl=+m
zcNW~w2T;qiDzNp3JDc;-gH7}+!HF?qbZ1!~+H}gXj)0!(ol6hVug4}+CB=zM!On-9
znY<4N)~5(3Y*nIVR#$M`*cnW-?f^}ix12WEe}Fj^O<d}3QKqhFfU{+6*u1`z_-@WH
zv_yFey@(3kZnT<O{xhOJF>Wk@&lC+7bCBgdmJSKah}p?Rz7H%z#j3Ys&r)eB9m=zS
z^Rv*zI0pL%8qrxShYKv?pPO?dxdd26ypOJ-@0Mw?)oPxsf_F=<$Z?{(Ilfcj7ex&H
z#mI_DChVQ}Jz{g)3C;xy;a++P*?u<}{T?fG=dI^q*7t4b6~?;-ts3wGcL&10^PO!z
zKi(I92t?L-+BbKqFwI?Ka7Ahc(YSLFeJ%88jYuOl4tvw8vG39N)>8B}P$BarM=*86
z5X#*ZVM-VJxx9K;q4;<Pw@;*j_~n!Ef&Imu`MG#~OAZ`Ox4^{(@@&L;8P>Y-C>K~o
z1aHM-;Ar+lmgW+~szfK!HzW2i+nIc4^NuEy{e27bA1{TQ*{gZi<Vt~adk`BAm`bBu
z+=LyErwJ{kGcog97gSG+gw3~9!2GNZi@v6SRmvT3mzZGQ3?Xj&C}1KR&4fyKt`f;K
zMF{&Kpi?UN?9V_oif)?7;72O@c-Ml1oh7T>a{v=ICSdA_8Ek0mSS-*M;F0CWNJ7jC
zVzT1}*O}9fdVLpAb<06C_b9`Ag8jH;nLdl1_Kx2zKE!nz_CzfIE+<&jfTc%6s6tyA
z&Uv4L5qd+oGB*cQU=GhqmSt+E$3XcFephE%k6wY&5L>yJro_3hnj5dMV&ntj)q$L@
z>vn<kW&S>#VMixd{p7r6{)PSn@#IXg8clr{BNXu>)sC-X@RHRuY_-hB#PTjYxcWC)
z5<HVCIJnb2ZUknMK0Zg{gWDI|(N3HFU{a<`pPBMp&n*%7*Y6nyeis#X@n@0aUwd{)
zZZ_MlXiYzb2=PzUZ?0rq0~8x~k%$$~NuGwJpwBM_f46-lG78<GxYq!`@vNh|<NCDf
zb0&Vcw2QOW)1zeq7jF7-OQs$?k9{ex#*@?p&s$Xx7Im6i=jBDe3~WVd*%mxF^D<^8
ztt2l}MDSK!CsAk)CvO%WqW`v@p?UoI(NWL<!A-@`(u^qm;}i^7-4yN~;%Aa}j;vDI
z2}_TRVckYg$m)bcH0L1CIFk8;o3zX@YIv*tRN+N*)Bc0c@+7E()P8}Z*;DMbyo80z
z_o5(iF(j?^qPjaavNG{zG+xqyX_6PAIrSCk;vLv?ei))_r8HYBo6QA=q+x~4f9T}p
zkCx-|N#hI&B)MbQp6(EI?oA^0&JwJp`YCUwUPNb~Dii)%n2UjX9r&Nk#ez2@>E5B;
zxFn0x^QP+f$UqvlDVd@7=O1L>gpu6vkCALFKNnw=zMYPXc7^m^$4Kuo3Fa=b2V08X
z@$S?%cv_c4Lff@y>mw_^&%X^zQs>a9(jhK(t`a>xU<R}AII<(sV)TdSJF=n6ka+LT
zC8ga5aLWz@rV;-YLJxIt=DJ_-PUk^5{^%@f@^{aVW=Fy3LNd6f?T7GOZ<=6QjQRpG
zrl{va)6cxe;M|Eot|XAI^kIxF_=?javM4w38C8d8(o+iw!+103y805U=JOo<v8w37
z&pC&#r9#3<XVzho%^7z8<=pp=W^K5Vs$bPXj^~rv-#kuhE(KH%o4H`==R^{=eFO2b
z+Cc4h)nn4zy*SwX5If$skajb38XYo|c@8(Av~3Cun5^QMZX7gt&VWVUvsshXebC?S
z1?|H#*d}##8YGiuUpfCDDqa1FACF{#Lf9D^>*2_(79B*5C-=EGN$-f?nTzPSBM?1L
zj$`qEYPqL=(U7#{E%tw$OjLg`Y!2HG9v1tc?T$4K`fP<(pFAN_?;j4z6{AVae;|MI
zE+#mbam5AB;ClBP?)eRL#mW?1xKjs&m!5!Tj2h?{zk!DBQ5crtEr_@!!Oe6F$H1eO
zg7EL<;J)N32AGbdJ&oqn>$5d&I$O*Qn(E;8`EJ;(IgKTLSi&>Q1mF;8!1BKG4)j`n
zhWy~qwAQ`NWaZFQ;7*9bvydB5?4rQNhqYq*6E!BR=)*M^R9St2HJ$oj4Sd(n#8t*e
zF{Hx{Hn^&=4`&B3>ibJBsI(DZ7tg`ULjuxtqZ?DT`UOS_ex&TQKR4@o6|NW>g%ODs
zLb>OoaDQ1AcFvf_u0`L*E8V4V?O6@fxMvCCl1~D6^%**h8^`BW)agE{Fw`G;0>Vbk
zqtDwdX!@!x^q1{FtY0(+`)bc~gEvcHVv7mfj&s1=FeUVV@&_LQ??oKG0bVkFD7{vd
zJwCe^L?Vxn)P7g4H2g9sZ|{Nu|JB^luQ}X>Fm1fSdD6&vVQ}`}LpWO-3RX+jprlF(
zxiaS=*?ugSGcKFOsx^G@Mf5WI8IPcz-3VqkTa|@Q$ic8vI)Yt0oY?ezwshwDG|o9I
z3ECG1Vzcf%Ru(sjMs0}Z+WgO<q)i*Q_^mdzxMIgzRhr29opE$in<Cr!Rh4O?A*RVj
z<B3VP@xo+j_Dt+J=}_D*T*30-%5Vs5Tj9psrzB%lcqUr&`$As@dHU+xXUzEA1+Ruh
zK%+*JYq@8Oony*T=S?`0-w~vD$7`Ya8Fg0Y*}=t_$zVrL0!jpx2~!jO=|yfO?Y0=f
z&#718>-a5LYWfy>A~LAgtKYb2QVuQ9;91oJI_%e^c+fnh%+Hv9art`Jh)KaeP?Zkn
zRE~~i9}@pVlcn`gqj?zT@Erw_1W&q<j$><=ZlP|*`C#+sFg^D&l}pieqSpVOlFuK{
zL&$e~HqrGv_`kGdV(zc8BtnI@%2o2-0Rtv`?F32WclVdB-NE_IpSc>h6*P3#Bu-Y+
zj5(gtr29(rna-dDE4ymQX-;XwoAbK4RpTY;>BV)>csCbfhDy-?fClV;@d#nsS#och
z5=LFxWS=zWA+D%ePG=tJhXbas(dp1pQl_#}Sf`{yv$$HGKbHxbk8RlbH#SsJp@g*G
z;5*L_J-AEfKCYsb#M9$2WNnztq9hIYtmH!SRxpc=Pae<SzCKQwogBti7omn?4!38Y
zC>wTFp!5Iir6Dga;}YHhnYw2Vn6-_hb7faSQTu!*`Qsy|^&G=YOL3}rbrP$av=1xY
zMc7F%c}(fg$6K#en9>fui*s&1?{3)v*F|-h;hIg94{|}}=0q-iEuW{|GmdJW%SO4(
zU=sM&22LD)gP6cOvS-F~T30P-)+Zxa7GsY~fam%A5v8?Htf<Jq1N(yH4EXy}7ow-n
zpt_$JUbNhYdw+z(`3hI+EF*{Rk)iPK_;j4=zXwYc{7H*qi_pof3QawxvkkAlqj2PL
zRMOXAtJ9ls?epFARg@L|l9r5<szsTGuMu^eyNxrpRz(-FQoKD@nYC|GLir1sC~8@Y
zPLASCcbx>SkUs^A2_vcQ*BIQ?V1mnU{D*7R6F@ssh5lGOinR~xVN%IiA{{o5`SeB6
zI2|wg=r5lcBVY0R&;e9C9snuIqgkeo7#(w2i%Ri5t<oZ8(x1_dqcnDc>zr{+b96i{
zT=p6h-Fx8(SIzqXGC=Ex7=2K&2`@yOpm|~%=N(l;aOn|@O-~>`Q!2nv7)(A4ioxl_
zlUeA(sZ4TQ45<vMM2-7)EUMK}=ukI}$y}X>uKd2~O2H>m#b@Em(k3#W!b&W?>5jg?
z#u2+``K0)>I5qN$g-v-I!TS1IdVKPKs8D`EV3%G8^FHr}pgp>D(&iO(gUVmX*)SdM
zyBnb9urD`GT%YxLsBxd>+=j%KN|-A#59t*CyEGvP12zj-p^hZmHjQ_Ic5J0B2mayT
z%Zq5omp%bEcmYdyuZ9y18l>u7BZ^-w#uJ5wg!Jr$#GMQ1n0K>Tuv|91pDe^J1&7eR
zaWrj77!tI8zXyd2b#cmX8J79dk)3<88_d*?64OtzEcccy3sp!3%^o9c&(g-(za~;Q
zHx*9&h#{N5{wt<-*2C=HDO9JRnJeqNLzJd(W`=X7aRxh1f$vip7I61G`EkOSJ6AA@
z_G<rwpQr3Wt9=6)HnpPsVq<1!cb1dhdKRxbT?I$|J6vR^1uI@P$o0Bd(3KfmamjQU
z`r=&<B&&rA`gU!EbDK4o!LDT}x7-Y-hwMVf2_IlF?}J-@=@JgU;s3UeB5W~j#?#3n
zd~dS{J+4N<w?LrN-0D$AqE6`W#uvX$)g&@C=ix@79~CJ+XD>%IsNG0DwD9sU_WH@Q
zHdRmDDI*J;j;Mh@eTZj&o73Jyd}m@o4CW_?WARFBy6I3EF8)$T4w(&L^o&qk{5_JB
zEpf)6=oOsB`!;fW$}y-53?#Gi%L!d1$;@O=a+iWcnDk3iQZ_l8#@qB@^^G~8DH=m}
zoNFT2Ege8I^(LBI6bc$|kK*Wlb6l4D8gJ}VrPB)?(Qj)qKKEORI!n}eR{kXtcve&x
zsN=yNZLY-6c2Q^*xD1Q^bD?&2J0u*qOuW_^q2$<^Z1-&o7<y(3CVM}k#20rwzeAqp
z_&C#)&3x|XEzf3JV}Kj`${>4@0$nrSjOAC2WB+g#ES1+`(uNn{v~V@`-KbA3E(_R)
zTqBr~Ue2{&oWh;Ec@cdM{ox+_PNd3e=b*Qs7#0Ylp}kTU^cVd_pP><S<@<-&`{xZR
z>?y?bpwk#Z1Zb?ON6oSh;PsSdEE}*Eez+mWhMr}j?uBoBZ^eLhPq~Pz{Ek3dMGnp%
z*~#_dR*ViVz_OAoVX?y^?z?swj&j(}T34ylr`wLgpVLja^@kd@b_iq(ANG?Io3!xq
z!w*<K_zzPwB&eNO2-kC}2S=VVKs(o6T>kyB)Oo0o{94+9{SkB6_(PQK8ec)We#)~9
zqi_80@nb(eY{rgLXCP*j1g-xz2m!(xn%uAhzaP(l9y<xz#=F~6`0U>!W!|62NiqLT
zBj_#H$3*q6H7j)9f@$CPQ4#Czs<I8G!od+2;o;OwYBMzj$Bv%D=5P0AY}5qiv(FZP
zsT<R@X~u$p{ADc2a|Qh{%^bZCNHJ1nN*nnz$mV^tVC{}m_;gkh&eN5p0j3(vU(pBL
z%u=CobsugySd5y_d&uE;ei+rFM2n~X6z*AH1sc6$=#d1#F~4P5V{$v!^;4B*j6Q)|
zn?&dvCrfx7AI4Uv4dJ5AZY<8$mjxEzfnF8@WpZjXYPpxN<x3Jbt0x9U0&0aZ?O9yl
z$VF)5>qcD!OIg&q48i!#KFrsf=SYO0;QrSqST@&&Jv%v`)m-Q2F(&6Q_{%<?8<Z<_
z-kAyeLOmI7n87+?zi`GuDoj2}3M2V^!n(=USQ=c%tqqxug<@aH?H@9XJS^o@`8)2F
zA|<$c-JIR^$OX0gnM8iN61I$Z%0&!4hDDQ<@j_lJN*)?bOvX2my5F;@?m`>z*t{4A
z%KyWNz9o?HItwR`dW<ImCa`Bx10?meDCx_!#!N96((qo7b@q&8J-Wqc#50N3C?28)
z6J%Kb6(Kk%ZsU{-7tqY=Gq|~U8=aDvPG4aY*<$Yq2kTd0qh&I2cD#>!o`-Q;v-EIC
zy9(q~Vz@U}38;C{RS<4IjIJ@)$?`NQu=|rtKL1Cs(|Q73HmV-xqzR#S-YJOcol3Or
zmZ6D74*ZNdf*v2lsh6z-%7zN?T4o44;x9+dZzW^ngpq7_y&g@!=msOK&1p6N&SW_M
z7JmH7fqDK%!M^W&=W_IPx>{F;rR~(g*6d{hBZ(D|6Q2*=HB*`UZ4+v>)`{Nw-iC1_
zZNXe_30r&n6M8?s$c6gN#2L5`S1#O&YPUY)wh5K=QTa`Xt~X$}G~`*qr|DR5;VP<}
z$bzI^6XxvwSU8uzd$fqgLG9i$qTm}RxR&FAqXvXnrxwR-PF_TXZE*tqmM)wWHIk)|
zo<YUae7Pi^6=naL!@qM!v25uc?wG?gT()5zGb`nDl1)cR-^Wo@{?1Nzd-X^bBvMae
zw+6tW<MD8Q-UMz%;z>*}Im9LXXT~(Qg0QCX75WH_$?&G7SiJcJ>Tmgs@uItE`jZR9
zaW!SJbUV2mQUr4b1bFd$43)Q)22BfAovo&V+jhoMJBwIUSdm0O&yT^=wK}vg<sT^y
zXILUF!H(r`gFQc0SpR}D{*J$rcD$G+n9+0>6?e8_iyYr|7iJ(ln+k=pM@ZS32RP7w
z6jluC)0%>KPBX%TC429L&U!uOB|!42h;dTQ`(b3e0u`RvPJ90S!A=ngy2x`J&q7e5
zFTN<?&2zn&B)tT6B`Az^^Q0?z|Blax3GC+4Kj_7GOh?2WhD~b{Ff4B>hMu3uW?#5X
zBDYOsOMbbdm--MU1n~3KT6a(#!}IjZdof(mRnUJuA2;>Qp^F>up{k-ORl9o*&+VE^
z?HA0$miz>v_%028rpxDf{v~5V&q|at3diX!BIt5^6MTIhMow@F)KuGtb(E?T8NMe{
zk}-hViVN_|u4&ZKMvT^V@g4TW+bExUfScGim2uyc*!k6$FeqDs$)!4DtvEx8yOkhv
zU@u&-8^t68G}w;?S@hT3a_&`E3Fs=hV)3e6R5g-gc12<Wv%D9uRp~J&<ee2!OBQhR
zV|Qc2cUP*Hu18gM9>Oh^QJmtQD#$N?&TZZ*%AB9hXNK-mDf5g+uj0#aqdtI*HbhSO
zwm8xtS@vh?511%F4l8A+(Obi>aD0j(TA$LUAy>QMP?SBD>(^xYyc<C)AQ(N(!$DWk
zg$f_;CtZWRVDzdF^>QuXne7ZFqNOEt_~b+!dy#wEUx?zv#$4w45ENkw)BcTb!cPs|
zaI}N(p!_O>NB?Egt^RvB)i-Ixy5tSK?ky+2(Z68p$C;d`>{DcCMA*vgCN$b6kGe%G
zu$+^|=HqrW#!r$Z@4bd4%V#r<Jwlj&LYHO9m4W8E$y})3H}supz^o06xujQT@zFAG
z(lp5(91|@m=BYBp9}D4~)hkZ?>SZD}xC43YE+{CyLC;b4v}O)oTe`-9dAHwzZDlny
zQS~9N#AwpIv4~syYagw2T!S+@wRmshJe+Q9&Fq%>K={>7^h2oyHr_I$*&pwrQQQQ2
z@?1Q)q>3;Z`jChon?%PdG{8z%EsQIZ<1=1e(4w!(`U@2yrA!%r=2)V&zzwanVmOc7
z#UOH_raCQ)=iIt)hX;FpLSC{tto5rw!?VsTG-DojuqO&GT;hEmYwek3qyzkU$>+}P
zd1v3eNo;<?5!^dF1#1GGSk3(f(50V@BD1$vH%{DCt<U4GvvXgO+b^Yg?qKq@KO62~
z_Y6&zzRrNwO-iAcy%gcYl0#gW*FiGT))X_>#c}iX%0NSUEpeNlfo*0|>_2aHX67Xh
z&aZZ%Nbv^2vGWgboYNni{a^#Wk-7oKzA@C$LxkJU^Bi>42{U}rgM&No;W*E57&KAD
z$Z9o+&aZ-Uc~_#LG>#;_QfC{FsnXwFld$no250STi(21a5XS@e;nne3+!RfI)3M|U
z{+Zo~+oifm*0>bx6gQ!pEB`>O<{Y+KXAU=n=CoA25O#I^MKWa_rp~#9-y<a1@6Z~w
zH_e0YVP~|TRYg|4SE0RWpM|R1H_*VhywhdnG^mUI&Yj!tOsk)b#;Uta@cr&o*38dh
zMf4F(DZeZEs?I*HK8l+&ZBVAZ0n>~pQpaT-oPAU!+MW(YV=e{_L7tgZd?6aTJGj2D
zvWz=AA1~jMr!9p`u;Tn>++i_;THl_Hr`^SwH7m#bje6{Fa3NKgtwdwf*V6qOci@!&
z2;PZxn9OZ-$7=IQEbA@r@Y>e^izd|J8!-pgAG(1)9Wr8q?ea9a_7~YUCJt9!mPhIO
z0^ZZ|1-##zp|(^T`tTXqv_1cEfx#~cPMA#}UcZAkha{<CZ!4}7lcs0>&+Ovl*oJl|
z`bqd5JA#)Bx7xoar9om;e}oeo+1`b-G@a>!yq8?y0)1i5?vd=<oy%x@R*VaMkb-R=
z-{H{hQPk%^BU(sBBXLOJqV=lqQ{zOoJ1QINj822ac1>ol^c#o1D*(Mwj=lW>^!5l#
z#$J7c+`doXu<@9^`KD(A-!@AOtcar#9-C;$_=O<yau+GE(BZN>Kj2c287x~a5aW5?
zOyPVd?Ad(_vv^lzl<VZ`x`^*k`Yi(m|IVVb`~<2kuFQ%X0lN&eSf#cEl?of;d3~4R
z&umjvc()OCJL@1M<t1tl7_!72OR#bMCUo%2#Jp9fN&C((s3Crt3o<)QOYcsiQsFac
zN1hTZAk}pGF%I99nXunWEn#d}HWsb)VAF$rsNE<vbhve|`sFJvkQyhBy%Of2`s*cT
zF7jkv2A-&Y(Sg=jei1e;Na31zPTGx0dOQ~^0><h$;MrCk^f=@tuwg^w)WH;JwkpTq
zW&B(({Jo$;VIB4PIGK)B)1w;lF+|E_G>drpg&Y4<2~RcD<Km%%Q1keo@XGxHK4)tJ
zKT`|%Iqn%$O8<%_{fBU8)*oCk_9AK&$S}=?v4Wzm$!wn8H?%(`!MGl6I#lM%bn1oJ
zxZNEUoi}3UHJ)n>)0oDQJ=|H7`81F(rC7f|07Uc`Y*UnD31?q&#ZNi>;K1{hzYm~P
zO(j`iWQraJj|HCQ2WjFpJ8U|mfq{mmg1nVQI8>#@M2<<Zq7UoY`DRz@f5nA1P0*l;
zgEsu!ehn*1x=8M>euul7uESl4iR?YU7pOgH%w(1ophKZP^On!YAFK9&iR2*9wfu{n
z0mr!4J4+z_b0xhtuNR|U&9qPXVnN?j4&%g{)!2RY5!&seoZ*N?D50W)hV`d$p!q7d
zU-Jr7EKY-W+rv@IQ=Hle@8Z;hk1*OHlj=WS$#d=8@%wH|s<v$&wSh4>cUmoojeduv
zYl@-9+>y1feI!_iG1Tu^I@WF709ie2sluvg;iuZkR3zjZG+aw2TfIW*^w-Cbc{1>j
ze9JR^#Hq{_z>VwHbJJ={!9IfL&kEl0d^{a`=f7gyYOBev_in-D@oo6p)Q2VBuOPef
z-7)NHGk0Y39=!QcKy8;XP_5C2{P0Qaa_(Ey2pt0*)L&?~ZV&OD`i3|!@0%tvTxRdH
zdH}Nd9z<DiJ2&Fn4itUZh>=@d*gElH^blz!j*Z5w$iWR8-W-5EN-`{UeJiS!jHj;m
zTOoA+1)kaX2us#E(#1yI=)3g~sqh=awsgFQ$cgb7n;3{^@*i-s?k8hn&};NId&pTe
zDzloSF@ol7doHfzH>_sO<i>GRTASO3+_91T3}QClpDoAMT02NMzLd0mOyUA<>@aJW
z0jvIe72g%=!@H6f#Kv|Bj1IHHBJ1NY<jP@xE#G&nmZA~$)wnw_0gcYe;yt%cGz{^C
z=4W@IrAwU?{b$EYgM#7b_zLXy(PGlG`oLF=XQCL)U~wPj(1Ss<Xy>7Hba2=Q_Io}H
zwkXQr*w0r<=Z+>EkT)U0H^=dvf*-;``ykpdCJJ0P%%;<i<Y4}rDJ)3i5_V=OvcwuM
z#%@_+XXtJ|L)(DqJfBU@=s6~o%E9EI$5?3|jZ<9TV4<`u)JHeMx$psK=lP=3{NKQp
zB0ewc63XQSd!eWFC7cMnUtpdKKJ4q^VyYDBj$%F|_9GisScy^**;APLwvgCa^L(*l
zS9ojsj6}6HVd9NRv}My>A$f9+%Jd(Ae-)Cfc3vMi=$+uZlP0vTAeq~`JQtc4OvL)>
z@x*+{7QFovcyBYG3o7E@rSmMf8u2M){w6W@)72mUy4-{fnZxjHVL9i?&u{Yi`M86@
zF&c2<HYfFGE>qJf$0^4Xz^Yc1w&?ER^djEk%BMG>>t`0#F?Oey3c7ek=o0qgh5?Ap
z%|?|GY1Cmjse0C1o?9^cAf~(9qS)UxbmZJwY|hdNkXrEy-^2vtFz;tbO`O0)VxQVC
zkKlI&o8E$`Y6N}j=R%v$hQMY?!rqCO;yB)Q=2zfEx7MCS!~WSUe~mV4l?;csC9PQ4
z_6GA;zD9pB2bich0+PIL60h}U%uZ64hB(I2-2**XuGWC>UDko-7Z)57o5FePPM{6V
z6VPX!5rxKooS{Sxr)KsO4aTg(AEQ5^^!Wtd4WxqGr{-e9?M>uaFoW<^-al{b%o^YF
zpY^|RX!gzEoh+)<EX{_7?ukUTb5)=j6%Tt1;=t#FH9F0?1s<2ASW|cv7jjUSg-;s6
ztjAu%<;xBV8eBGjMwu4PuRD$<DUZ-k;x2q}JB;H%g>9EvO+&pTuqmn(t;hf3;<p>H
z((@4{$A-fbLo<Qt@Ql67X57)`qp`dq9?$MEXOn-NLS>ERNEf_D54R#N^o=a*mKwt{
zORf<`_dzWCSwr++B$A=Ihv3`iGo*dXVM1OcancE&QSO*F<6}mczN-jL_MXL==f<(C
zN7LZ+@ic1xJq{!CrC>};CcHm-3ja>|0peqX#3p_x<a5Wc!G&SZltEZ(I~m7+pUYND
z%QC-0P5ka9M|IXLVv19|X{O^p_~dn&lu6d3i0<@hFWPkB{Js)m{?>p+UtYwbUN7Z}
zrk({Wb4%{bFEf_G?^YjrYO|Ptb$D`vyYR)D8-nVX;~0PEF?YQv8?=Ij(E8yDCWc-i
zMeBk7Sh$@?`{si9YAF!dK1PKo;5N;EjS2kcJG1OE+8rFtZQ`?`gGSDfH{l*R{g;2f
zk9>=xOU3ZSjTKZgG6R=>w848{b8uArXnF<jft#@y>eVuIo~KLKtRBS;=L#4*uf$Fs
zk;FjhJVDu=ujs|x*tZ1$<`q=1JYgo43sGhT%A;xH;8YU2lFyS@SW!uy9bhu^gJ2K8
zb6EDR6|WvXg{=?uXvt_@s7({0&5?^yKXpA_JG36XvfHuNOolC#*$=&st;wmf0IIg{
z40xZErJ@_PNN)wdD?9QM=ebN~dH&8!`p<jtF<nZ(`-(HMxLXiBOt7`&I=N5~h2K2V
zP;B{qSf7_cN9@ZJO4qJJ=fHo&DpQX|_lZ*0p~Ye|dFJKTC+O#tf_@4Hv?OB*#)^-I
z1y|C6EU=+#jO6I2CJ`1PoGBQvN`%j8^SLLjrZlj@jus93u;h<7;rh&an0ibVKOCjp
zIxz!AN`@eRNf0i*n8Ew%CKHitMSlN$A0G~8L*&&;T=~fg+ZH*K@ne^<`<HfL^TsnI
ze9sl=_!p01BLx^Z^QC=*<4&wRJ|8SorRaN+3#h@*9?mp;2e${dY__E?J76=2v!^~r
zo0X&Zd*>~<2Ub{>{RQHGokAz=ji^+1mAmQ|OJjRpVab*m{QjbcC?((G3TxcCj=6g9
zb(to}%y1Ao&3;6rix!~x&<R|0#GKmNuA%J82v|6M2X-p9g6p$8*n7I1+pzc=Jko0=
z-3yiYoyBO_aPm6(@;-+vqIzuCR3FaMz5?7V=ToDmG5Acc87(iE;b=7x_UVKIWux7w
z@99bW9yL_(p9t@B`6|V{-mFLY^n<kWWdi*aG?iI>wPfR8#AB5EB39e!2VtUV0>K6y
z9Ikbt&ez(2u6oZsxh8`m{Qb1b+m-upb|30DM8e%MJ1{7czgIu+#KypJaPQGGQYZNl
z&)8Oje76CeesVqk-N=Tea@F{!c{-CaG3O2iXF$V>=cq5NfH|^PaE9?1?&Tp%ObAS%
za;JE{focne-R0f74@I%p>lPZ<-$UoTQ0xwU3a^h`KxL(F{POB1|Gr_)b_+`|b9*li
zwY|lE5n31!e3shzkD$2|tw1JF7rI`3#Ox&uUGK=yPJL7QjQy{$a{;UATibpIl1e9?
zRgy|4Nu|=7&ybKPNs<s12?<F~A*twqB&S443Z+Q-JG90#gi1&y2?-$yNyz#14f}n+
z_uYGc-@g9yy5`kdtF^|QYprq3HRto(_wSxdXY0?vyL~%QEr=fvFX$4h^hBaup+=p%
z6tTpZ7lqs2f>TC6=I}%xAIFd9{n0EMy3Uc9s#}v;q1R|d!L>32hex#M)F`%a$~s(Q
z=L1@=%4u3c1CD>FfZa|<;KE^Nyse&z5ksmmZ~rZ*p6-Lkb9X?&(i8Cd%yX<*+k>fh
z<VoV#C+KjoH=Fk52rM0N6-RC=fgRQMXxCIA49`v$rqpzyk=H3)aJ&?a!hWJuYy$Co
z(SpYL-8_#iiYOQ##;{SG0pNIBkhL)<hC6>_%#=ZRRq8uv@r=v^2ldENEolsE=CwSF
zdMM7IqWCpW@ypkFtRf?sXxxj12fyZ%Lk9iWyIg@4mRkx7$~Vz?oh4wtKLNFzWFWoY
zX<8dE#=4dRxZOC3O1z(rEghp#dfHT?H)#bKylXCMomq)mrspv@@Fl)38%jRyF=k#{
zwZ&Q6LZMt^wJ`IsCD~XcU@y)&JKUDd^%bMZ7msngf4B_F6&yi7J5!uK^$XoT{wR!V
z>qR=IxWTmh%dy&d7_+a>5t_rwvAmrtP+fmQooW@j$ZSHNLAfM5tw22a@-LKftq>$L
zwJ_^m9W{&YO|F+qLdNYZ>hXFioSi>_8CWKZKikTZ?wh8VJ7*nEP%gsAo(@c$^j&Q9
z;0@hAg8=RO8+-Ee@$$J>xYYbJYO45>>i$OTkV+W-YOg~t)rT0pDh#?^QeeVkPcrJ@
zM{J!lnH+LhPn3+x=!eC;_bsc-W*<*Mb?%XW-I9sri9VdO>5Yjc3jAC*0Y2BJ3(@Vj
zg;sePq4wfrw0X80A6;CHqdY&N>5*kZra>mAnrcw`q#nwid*Zz`6*^J=C!FT}eoLh^
zsQL01Qq4{}?Qh#jc3UifU4cW%@oQn=zM@G;Y_TU+b^|a)Plw1B%_9Rj7cp)09bq#Y
zh%v(smA%{Igiqb>zy~X?_n0Or1ix{|N52l@6YUg?I%0=e{zcRy!i0Tu4iXO5K0~GR
z`jFb;%bZVc#F!(kW&T!LOg4H4*L5_|rA32T%fq2~{_YoXxz`k`+!aQ<H9N&;)vHAd
zCk??>KAN!q_9^Vsr9#76r$KJGHVcl_VMo1x;<6S=_N#RroUV>WZ*>C7r6cfc<sdS8
zB_Xr=0k(3Mg1BlSOV-drbJ=5@o#90*GkUSy?3-wpl}EDjXTYgz&TLC=Z(OZ*6|yvo
zNX$)FaRKMK4zF4dq537b^5K4>^=dH;SzHJaS|W6v(S?e4K0=qP8m!r)MpE9d7Uue#
zrn{46*vaV!QB%r}NlecN_IemwA-@A-iuyX8pM9B58)C=wMzo`~^$9xFB#s`n$;G|r
zYC(u^5?WW*;_DkHa9OGjX)RbK9z3H8_T&e^#5*yN)u_T=-yec=W*>!$P&wwHA5PQ{
zS`g(AyHKVk7acXKa7Cpy&&Av%wuwHEGY)HzCgC_PTNs2XhaRJ+s}ptR4E?3@TmwE-
zgnOedP}$jiVcRDKOv_SZjg!jo_u8T8IP-%z_L>f>o;4bYHrI}Co8T1x-V&s3+n|0)
z9oF`rj|&eDC9+@p3X@wmk&=;*Veb>J?@$XS%F9jY?Mss&U8NthiD`gi9yw%qktxBa
zvM{!JAPH$Rz?WWcp|P$X-YLn&jvIB5dwVu#;7zCM+v14E*@<|!a}znaYzd6jcNKx@
zUdWR(CDTR@V={xI!MAx8?r5=RIXqKq!h>?$5c(aw8Wr$Bf(n)^9Tp|8Jr3#}qhKy)
z(cHF?AWHN-`XA_zPb{<vXeZ%~6DdNEaT-K#F$WtLC7g6*HL(iqL)xVpF}TT<1bufQ
z+a6{KWpz@d<qXfvYh6U7g6xF&W)0@@+6-cQEx~;f$Iz#&7uz{ij)ZlqksJkq-K{P}
z<$L-VRpH5A-7v$7=a=DJ)-+bMbE43yuPyd78^Deq%f<a044{0XyjZs~TZqe%U=O1v
zfyITh;H0+|*T?lo$H!`ntji;puV10b^G;x`t2Upb(lqL?S;TnzFk)i=g=$K2zQ30q
zC?1o=_FETGm^BOCRD(#~_0xE!UmA3uaU(I__faNgIcLD!6uthYN+voMfow%4n*F|t
zCp%==k@<gN)3+iBU3mbDhfhXDy9)fk{aD3=YQ;IDo<Po2X>xIZG0Fxs()Fol#cI3L
z(SG=EJhR@AG?x~E-_Khpt8)eGC-jEz8*{LHa}w2wOeGmo<FGMnANWn(hUpsoJa$(N
zJ$P<7-n{Sy-|wUtR#Ph2#i%m1FH137MV~pJP$DBWqVb+w7#{s`1m^Ipp0+h(A#}uM
z%#wIZg@fN{ZT2RR&^QmBj&azu&4|exBx8%*Xkwv1i;UoGw$-w65b#}{`y(FUxyTI|
zysMDvlv_iSQaI^<C=EX)kA?TodmzfNk^WdVh<R1Lp>7^E5N<gF!#{Vz%j1%)ZL2!=
z9$5gTt7l`)oAo$3ay3&N*bMI1YtjDFAY$3)5#6{(9jdeHF~t5Zwd>R(s_WHgkNJ4!
zq1PZdp0g#7BjRaV<|*VAVyrdOWj6E+op5;*9+JI|S8j|Yoi8uI>`?=Wf2Rg(9Zn$A
z;R>GoeiNHcs}oayZGL|4jsusx#O`cW;!>RjX?xS?{W05N&Imnb`|b<&jE$zzJfLEq
z=2OsKkPUgCQ-${F#;ob-Kt?yzLNvWZCBl1&&S%F_pJ^xIsI4TtIVeqHYRgbsN0JDh
zH}RYPbS(C^X0Ce`;Gp&~n60eNitqMk-uXHBDZ7=DN&YC|%X4-6-Gnfisi2?#9%qkx
z0XBCpbG==hxX8u{%{Rv4pgk(ABJViV_WC5;NUg+p>n7U$ehYq@6axLM$I-wfb79<a
zd-O~@Nqpp=qjJbH;xm3QyRT3W7iaT1H{?rM#G~stdYVAC^i6=k3I#^YG66pgW!lb~
z%t*1GmK-SsKc>r!I7d-@CX|d@_!t9bsS%qjUiWI$g8pSDEX1S~H$8J^1z~4!O{o<N
zEUy$v%=Nku=FQ+nb33X!T);>E#^mUH{{Kq>ETWekglloez29kUdEkPINzKq*G*^7-
zwj25M`vIo@CC!q4JjB}EdMq_q1s2?^d2Ie>beVY!H*;^#-8=QPHA7a22z>*hN6A?5
z$^t_Z)3IbnC#@OPCK#qpWipyuF=lT6vg8*#h%D(%>?D7qwPqh-#-T#|FgFF>>y}~8
z)AgwEy^$`dG)1+{4`3Ia2cpeULaOak(O#~Z9yoI{Ib$`P$b6RMKI#e@Z|_K&=fq*~
z><kEUI*ElSBIAeG(>j)lC6=Fs>-*#Jsm5xU;sRJRaXZF{l0ioAG*0~W6Fs}L1yi@T
z`1E53&M{2p*CtMAvc(B>RvV+C#z9ce^WfRvV@cEfPFyr_Hd))=gKPXXSg-mi*uCMD
zI9vUy$nCZs{;Hb@bp`8)r;aOp+P;(LvpA80pHo<2m_4){IR(eV#?e-uAJ0n?LSCLO
ztL%~>QI<d9dZiH&9^V1iQI9~S!-uMkDHXgl=ZGE-xdb&P??h&?^=KbBin`S-f$f?*
zU{!V<JoM;9z0@t}qEL;WFcmXQeTmz4Gp67*m>l2cN;9TTMjgtT#>?x`Y~MR-*!wYD
zipWQ`IeqDE<0+(V*l`HzQweX!Ct~Y5#Onq<sA}m%RvjD%iP5S;XX;xadbb`c?vY`R
zyKSj{<t^A}RSAxj5ws(?n)(^M68B;`I3{8jgwD`sF4guBkk*&XEZl&COf~L}G{$o7
ze+)gn59Oryqn&&%b^72S9N2k^h6lvZUo-1ySdJN!;q3h%-ZrG)iMJ3)8t_eBHu~hh
zLAS^b3~Zby>=0eYijObEH}A&@XMVN{27~h<-smOTd{4*g1^L3MU=ggm!t>N_=zzzb
zO5th$2~4+VCm2PIWLwf>v8-qaMvSY2eBWPmX9lms2ugVL@-;djTbJmkXX2M)Gk!gP
z4@Vx1!DhCXm>T}X`x+i3yYjUt_3UmTL#{t7|Dj3`j_A#1=NmF}?K3F#Sqgrs@51qu
zl)=aK4@Z9(*z)?`>VceF9kPG~ydKWB*=<3YDG^v>Hb5}$f1cNRJXv@C3GsHz&BT-R
z7Lp$x62g=BU?iMEhtUIZN}4$?GUyM*mj=^s;xA~`n|rvg?;_sMW`o<iK$Q4-2ugT{
z=$AcxaQ*-}wnJezMr>4s06!xZkolb-CtXR>qI&din2SPb3Umigpnm$BMXPG%iORTN
z^kf>(j$OA*$l1&L91%vMK)F=$I~6r{{*D;Nhqltbhpyw$4lO*I$8|01bV=K#1tizs
z4W)-@vubN^s4egX>$i^LpJh{+a>6KVbYa+2zJlss_Jg>!iliaFg_~`+k)}_hs7yb5
zGIypslk&X^8<y`ukBHs4&HXGcPsoAvpRHJ&KatK?UWRX`Y#~W!(lC1EG0+)#ocB(R
z;FX^$RvY!AhicP_e%)Pu&Jv7HnV*C&qqf3o=N9<VG7s-Q3?pjRuFPYa8V)T~go4c$
z7!k)=@P~EDp*r5%d$Sw6?SMtV1XBHSHKERv*!ku2v5K=X4Y(J42YO<}-UYNPLXDkp
zE<#y76+!)BH%tuD;dR|fP;xm3v;}3RWJ{P%ss=MST?5UxZ=>Z0Gk95113E^7Fgvnd
zeB)jj7GHdX9kP>AdhtfW4#;Ci3k7}eI<B?73*VJSFcsAhT%S6F?odj>Mc*%?RpDat
zR(CSYb>f_<dk%Q)!&OX4G+_I}f=PoV_a+WaLAo=5nHcWIrxuxXY3ETPdj$u8-P2;F
z1FVUM#BO2s>Qr#PobFWG_X_k&C`H#5cR+d19DMXtlOztw#I|3JLY__lNuQK}t0v7N
zO*_58YvU2RJE{w+f;Yg-G66S-X)(V}P0Sf-$4>ITg<O?9v*0!7A7KudT0EV;F_&T+
zAIOrpMUuqX*Ae=DKTTd$YLS`8`1M&yDpuJy2{ZTWphxTlTz8do*rS$0Pw7W&f3gKW
z4eQOU=V}TY=Oy5n$qR^PQ5>mND}yn|8TzOH#NJn4VoJhQEa7?DTI~a<rbQv#eYzdj
z{=9@WS9Mrc^${ZU`av&s_F|!0?!<KB3Am`050=_v#1`|oKbH3iJ+ge5Qr>17_A(as
zv>6iLD|%$o9^Ma_TrFx}zXDbzt3rIQ>4HLN57<j75lcTi<`E*veA>Uz14++?@*sW=
z7JE{>vC{{--PZE!ISG7HIz&*t#(S9_1G&%T9A<48M0)><fvCi@uyu+ho-5})h^6Py
z<>7YfDYqNflbK{&<tW@L#ra!#g*Y(oHFm8%$opUJ*msUMyxe^NJwhx%((MXo#y=Ho
zh8d7?S&nRrhbi;Ex?kx0=7GP?egvP=$&74qBX6|&uo?NouyEreXj`gGOy%nZ`|yD{
zt$L5p>bwuHya>kJ^lf-=-6L^g>3HI9E<v=?^oi6+4>;5Nf{>8+6+hjdLczTW(wp5`
zMnWHUy`L-e<~;CIJ+lS!sStG5HsM|4d>ozL2kk#7l0-RqW|(VBmUTV^>kB^AC}t<v
zS;rHbQ)#f#iff1Fh&YdVI4DmvX6_&21&0)EHfzyk*xw<`ymW&h^4B!B_Ul%-Hc6SK
zB+X#uwzW=*4w9_KMw;xp^$Xtjs1wPPw(L>?=kt}9(z4X0q*1>aBqZ|Bc4(hOquUQe
z#fK1;7VX6Peka9;uNjbY&*E`Lt2)zfP=b>4&O{+-2(*1`z!k-(@YGu=R^=N-4)h*_
zHZP82ZnQ6$Dwm)S&s0nLXuy`PN+VC|rSaCjUQGAd65@Hvlw}P-EUtPc_$+=4#lx-P
z(yrCy$J&8JE_^XDTlE<CSRVzm><z-zo-q)3D3nSDEkoUjNt7nMqpdoXu<iXFOjP=f
zM@v88NoVe1`}J0w8)d-sHe0cd$))%!=ratek>K9F0pP%M?Nio#gVX^_+4oD*@P5@!
zd?;bV)=!_pStpN$`-aybeduD6muF6}w*gBWbPeibH^Qs7&Dir?g}m2nf$FK#g+WQj
zP;P~Ygx3`a^+UhItFf<e$;NRc_r3~qOwb{jw@%^cZz5P*6o<Z(Vu{bdbNJ)?E1q>y
zD70wGv8CPeSXAmx^ryOmRkIzD)$+l|+}rZ8IT72e$6&<|t~V1nD?iK%3R;6vRFgxB
zz8aI>E-Ub$<TyIkwF$aywMmF&DzU#JPrm(-W6M0YVe3XJd|O<I{=Y8LvkwCyHE9w#
z`!yBnT@_(^o;w@hp~ZcS^TC8?BvdHuq27-eIx6kPm>wtEV1EHkj0iEb?vI6^kMmrK
zuNWTa$R63rFmtXKe(}+X)rREJ<2DOO!FpG$b2oz8jW6L6XQ~~GETTT&jaj3@0gO@3
zLkZpkv>bMb4B%eS+cNuvv1Z9c^1TnM{k8^ktw-^m@Lv~XKp)Q=<wD1ooj9xO5q`T-
z3ptm<@zP)1WA3U>np4WLbjuVluo^8cu*=5)Hjym8?*h8FZ;9X}_vil=fO*|<s4V8!
zIJJf7cKQeU>@;N!0m)R?j0(C9VptM3o(vhx^W!hSgkcsd@#Bv@fYU;8xvK%|`^+BQ
z&+zlK`||8a^b5#c-a=!q#o@i}Mesc>01n@|jl09#h~<=a2zYyt)Mur0%I!Mfa9k>t
z9E%w?C#giTWm!?C5{$mV*_weFFx=(<#NYQO6(J|k;LvAmRk{y8j=rSGatT_?DzLl9
z`opWviL7=M_b7ILhl#^Zk;1H3)bCWSC@1<FOjycu&@1K9efj_vazzo`2UTLoihKBE
zpEV}51>rY`o#?eC6H`_WVrwIYvy~fP!L5Y%s2toV%pB;(Jh(=r%SDQr&QF7`(n*{V
zl8u`KP1!{JCEOc!OxW>F6EyAaizANP6UCC<v?r#UMtoX_qthgDL47DFjA6pi&>2vz
zn=KmfJBWODutLc<v!TO~(!?Gv;4Cs{RqvfK%d87;mTu*I67E}RE618y7qRM231Kh$
zu>8pqxO?$1_U2bIYE%`_AU73Qyvm+sU%X07CcUKtx-5v$k;{)Q>9l{TBTH&HL0Wj7
zMVIFXKRQZrc-T}Dk(CHOUGv!NhF}tBVL?U>vLwMT_oEMI=Xh_|Wh%FCfoZ-ZX?-B!
zlpxoilv+H4li!YFU|?U=bQ;W3HgeWe&^z$IYKyf=gV@BldZ>Bs4lh$I*u|$dn7#5v
zS>G*O6Z<ZjNZ$mexMDc2nz|LoZR&tY;>kRZ_AMkVR3Z9#=P+G$6E)*G5!JmW(6$ZJ
zY2mP&yjE|5jX{Pa{*g11J60q<V<ooieT`UP#SVThf!dxB+<f#h=PN0b^}kG+W%hP5
zI%Yjo1jxgLI7J8!AI>E7FAIshuenLX8AobHgUi}8yuM{mE<Ab-C7qk`ub436@|I%u
zTNmtTTZ_62q%i1XEYTU|2U+Ub_;mAp>=&Ai@t-#dzSh@q&2)9reJoql6l_Bx6cD{C
zxxYoH6}4WR!p7Lkm|hY}SGt5^J*Hq)PXZ*b8_y2y<32Xqr{MfUN6>b!!7*d{v&dbG
z@aQlz6w=SZrjS@H{FINj))TR=V7vIm{f|)i21uKTb7eaZLkH(yTprK}jxnmF*v0~L
z_Pj&y`DSq4S03(ejKa6Yyf?4HnWF7l0+BG{Y^V}^P&@~^Haw=ShosnT`P1krT?&m?
zB#FYjQFLjhDhnK_Dh}MH$u=F-V;T?qNp|!`nzLVy#hf^O!8)K1uC0<G!KMXVM?)bn
zwTs%#{{;3PN+ilRLlk>?39k7d$F#U#!14SDT3z&63{mMqao}du42naULn0EVEz6GS
zaIQI1!SL04;Jm{+lGL*YjCc-I9P#Ix&Lg<Q+aL6ynrpY$<8K)s_A0Xgaz|-HftCtu
z6my1wXDP~<I1s~iB2uiEDl|S6!>U6!gf9jkIVX1>b+g?sHlNak*ZQ1;w)(5+Jv0p#
z%uHZ2{f`QIopR)I2KW7cJXIDXE(e3|e#~y>G-iLphAsQL4fg*Q!2DY?J}^ubhALl%
z#m{<h3f`dE-mk<u4JBB1JO(n(>98wv7m)R$SV1CZ@`c{RHQCIW-N+N+@mW>@9lmKK
z&auejo@g~}cg~|1uC(Fl=_@g_?FDv?bYLsprI=awF=&}LhAdoRjo&8gk;r+bEaeUN
z8}?TuLtYNT6KZ>~%2)*5v-gR&ROsP(uLXkN3`4NlG>FV|mO<n8b@(G(o`rqvq3->*
zVRJ(T8kj`EQE6Q?^e)3#$u=Q;NQZdNy&UwvU_#zq@B+WlU*QAiz}a1nCpj(Wp<&!Q
z>@2(txtq3;HgX9M7#0%Maw*XK6pLT@e2=)LMm$!;k*d@7?2`0Q{IYP9=pl4sLQD%t
z$yVa*tQcH!Fb$IKxsk9#F<AH6mbsi)V~4E*vEiNxnS65-3FORVg;NZ#dW%@9k{*Wa
zpF~PlAX-LVp}oq6z}II!s5l~$mMq#%qN62Q?kp8{((wvdy>}o>?Dm58hZwvwONNaZ
zq{Na$`grNpWvme|BXWO9lj|qnp~ChOG%#=#=a^GWc)-0%RThXL7r{z)6RtAK=lsn1
z=x0^P`G0b3;3x%lCf<r9en^GdNyRuYn(yCREpdu(4Hf2mfZFLh(0E~Q_D$wA&Y2v5
zbKFfZctVk|QEM2449bG7XNMBEFSZ!7Ntyjhc!q*}53OqK&*Fa+Qm@5xNO1FfNVnj%
z^4l+|;bBC&rb8rgk6%*`*Jnn@bXls%n!Rz6U@x82G4RR}@euz!oDu#W-^ll3QHC-w
zeQPXOzTb<sBN}jku{EPye{K-oBYtgPk2#!SFtgbe(>*Q4$3C7AJlgi)9IxK^>Rm7<
zR%nRl^h2IWxf}x}AB*3;Oa?z`YnDCTizJ0@LMQM8c5WY8l<SX1g|(2O!sjJbW<6%s
zf^7Ia7&<GS42$amsr9^vyx5D`Icu|Ch3QDlw+bI!oAC3^Ahuxx&m$=|qe==(*u}^e
z>YSHIQ)h|5p{1Ml?+O-uNf!}`zL`8*z=?#Ne2f~*AIho+pfC4_9S`SOr%BVvoC+th
z)c-Ois{3Jlq$_zH@f%+Ms^r;SoJV+iGHbFStp0Wh4e`+-@?R@pi}^s(JF*|sUUCM<
z)ec3IiJoNZufyoqG=pR)C35W$z~qxEu+*Ul!X5|H6umuochxqcKJhqiZI>adx8KB7
zpMF!nZB{JR#SFnQi3ES_B|OjLdceSVywS<^+@?I+?{)!gzWW{1*1y9g*$ogCHx*ll
zbfAv?Z}FEw?#x<hB)fjNP4Ef#N7JJF7!=C=zsobhJMb<(IQ|gnqpLXaf(5$~nFi-V
z{qbOArqCx^k8E#J<JbCz!osP^xaz=LTGDZgRvqa950lX}WNWn$cVhur#(ky>hFN2l
z=K-p4suYz%XA{4TqiIk3W?FsXBW*r72`6-&K*!1iSiL3!AI;atj7!Pne0d@$HkzWu
z?lL-q^H00Cx6(H`)-Z0tUJ`1w02>ZpfCxnoD2FO+RM~+;?s854LN%Ct*_FM!na=C-
zbHMTASgP-=hI60E@l3Dp!qYpBq``pKqE&uVmqqSaePcKad0j0S{B;_XbyA5z_AsJe
zUn=ZRDiJ=e<JytB#iVNaLUPnxl^tzdNE%Gu;Mx&M!j&UGx%%KJuido^zP2k-Hr0*H
z-Taw*abM!F0b5CY#bW%SEkTTycF^|J-}tHWJdO)Hit0Pmm|8&`X377B;al&Ztk-)S
z;=35?=Pd_oE$*p7Beur;tH`<Al|^yJUUfo%kRGGPq*fH-j0bZ;LPfr;J97erMEOFt
z%xv*UEg5F%B+s&y7UN^i&5a)07vnQb$m(zrrpztE{*umkueK2n4@sfMM-*AZ%)ab=
z$PK)|w*gut`!NX}6z*G>QmxZh#BRKIz1lyP6h_Ijoj*0PAwU!7D9t62j>Fh-byw;-
z%Y>DNA3*&_*?jKgps%zsYtFezbI-kicU;@Oy+(^X+OeAGy;ub;2_6{q*@1<q7lG6&
z?(?#hVyD*l!nqA*Y;@EBrmyKqZk(lHFy=CBNawwPuQyTZr7|~&D3f=jKQ_Fl7?Zsp
zB^(wqle2ShNLd1^9F+iXx!-8uds_6dU_P{07!ik;uK4<lA}&=(C5DFc@OjJt<~{f~
zp8muc)!CkC%(;(EZQ&5e^;1!L{;2iCmTONrJ7wBH&Mj)da*?)pT&p%a+@Q#2r64v-
z_9j$X5>pL%&iH*xvUXZ8*0tyZgq^sG!{`T^&BltVJ?l`}>IFJZ_a<(Kbdjka#glVW
z$>Q2}p||B0xUuFYHg4Gl`4dL5rt0^CgN8jBnQ;QoULVY+)uhnIpglB6BN25hJMd=J
zPmH=cnN$^P!m6+7DE7|9^9eqJdr2dnkxwS;O%9_MOTfy2v8<toXUDv#0Gaw)5!Jm1
zC1-bmOP(eyoNyd0=jhX&5h3_Ueh+pfkA}wlB+e<X=f7{m1l=Qj7*rIa*Y!xzVE6s#
z9B3luakQ}W!6WGF*pJ*&xQhz+R|&~MFF+$Oi&TbrqS^ZpG`D^ylrD7ysqfL~7$+e}
z6i+%o@t!}*SG!<E%o{r9hloW#GlIyni}cA$N4PO&6PYLe3g(J>tUT<vQ)_Agj-}Jt
z&w#D4%S08^`8mX;Co-`8P9>F)Twm5l>L%{ZsX_nVE5K-(D&2AF34Grp!vcj1w5<Im
zyh?1KKk=)$TqMCt*P63Cdq=V7#}0z^z%J2X<?+n!&s^L$3%%tFK``}0Bd11TEWE&$
zV<+&;9CJ~~voXxHz>`$14nx1M^YG_yD|9w*C_6IYEl6+uhBbfX(|#GxVbe}M7JPww
zwmHKZet3{9;=>NF*F~jas;vHgm1uwMHoP{A^9U^xgjL%wLemKwmS)d==e5CjaOVs0
z^BzAepF0Zbb*u#7zYDp=i@2^b2mOv`3JV_YfmtqF1i_y3Z^sGH*F>F-n*Lhw-pAP?
z6|zF#N80Ri##-THuWalxkH*G8Z}N8IEPPoTL>gqY!Dz2Cy2L4yge?zIX0|nJwVgq`
zhTg%R#b*Sa&I0uDI7t<Z71^QH*RWHg58Jr!lQ3~YI~Cq>-IY}&2A)a~zfUj4xYi@s
zttbe!trDVd7C^o);m3jgXGMyc8;G8}Dv?Rp2ip!B(x{?bJh7parYm2il~*i?nfEBt
zOHPK7%1o;EPK?`PmEh&c9kfL)6~0ZZ!%fGXp}g-l+*f-5qayyo495+mU`mnj(6114
zS0o6HcGASt>nx5C`eOHTbG*g<C%LcsLeA+v%ys`6w2ZA2r}f)KbhccfP2VpG%DbM6
z8=l*-$WPM=9@~zy)r#Tf7FA{*@Q8ZdKP_sQ?9KMlzW`F#k@JsdQui1WR^=+q3U-OH
zwZKXoujMRyn`cKf7rUTj`%rdxWGWVL=Es_-6N3KfWw^&^4`eyZvO|w25)F;*=&PHH
zB`=qed-aUQ)omgh{fcl)sv^snSq6Pq_^>#sN9cWQ5(|90lv-BmvTw!;Xy1$T6pOQ+
zlz+!V;sy^KS(=4S+=F#`>;W?R^+FuLc@@di7Nd20j#w^wFq8P%gT58}$*eW8LP>Zf
zwAQ;(ORg(c+B%RVt2fixiQAF7g@fmn6g;tXAnWqrzppb+W6GUaP^;no_>vJqtBo5A
z?%fC}M}MQ?m2h0duhXI|-wADLQdH4u06P`zhxIpm#O3w7;o_zF7(STi7V>(l>Lqz7
zNE*z{i{_F;oMTrP@s%#Tcoh>*OOg#2(_lu)I(S^O8P`5FA}#AA$y1ka@Hlx5b=0<G
zdhVfYEAGG?&XB6xKZ?DK^9HwNA-HYw4e*;bf^_oQ`z@1W=y`b|Ip1kYzBtI@h39wh
zIL~&t^+OeJ4Qi&V6OIZ4rt^Ny@^OOw*oRcdb^`tJK#gvyv4_=@zR<B-(gd4khO?CK
z3H~*Mv15oWv7daBN)6hDQJY<u;Ur0NEG_~hZjCCl_dW@)eih^JQ_rz{$4aqVdp#c2
z>*0yEk(_gM8=ZguhT=ud!Y8T6AXN2X@k<6VjSHNcEjJd^hh~WCU0*=5oi1iaD~k`t
zPr!Y-YS`311gf?sL6FolG-y76iv~Y}tvz}uwcHr<2YX}KqA}!o^E&E0_aeQ1>KgTK
zxPs9MGAuhh&q>1fhLhZ#apcS^SG4!+#_?WN;1T;4zA*#PU1TIKzcUd>pWP31j2aZU
zG@+5AAvu_?!oD;Hk$Dvn#PIWQR*~h6O(G{|DJg*(``=^ciUy3mc@yJvnqj>~IcN=D
zO`5MSB}tz-KV4xDnE2esoLBF_%e{~6Mn9)HsnXICe^xJd>0S~Nva%BY&tf8>D-jtH
zv~<O?uvPOT0v1I?%wMoFEGQx%a(=+Fm672qBZu+FS1ek#)YU~xV&;EXY~}uJ@mAJ?
zX;OR(f7buJ#Ssz%{%(s_!9kHh!&c0XSQHetC@LtDZ#`g9u&ayaf7*fJHV%#>9EaPF
zw6nJ#Zf9#V!ro@MwY9a~NZS!3MpzHGv$q>*W9?w$@MkZ#eP7jN%J=Zk>g6u^XJ`NC
zlNr4{M&TI0|7RWY=O6#F{LSa|XQ!>g7l-}(py%GU#7Xjj{;XF2GN1#0=x+wK3RxDh
zBq%cAzrTckAA2)A)0W|5{#pOLq5oZO{r%W;*GA4?77;u@;@?O8=c!}-zeDm{zWn3*
zn?3%=cW3?YL;myJF!`4u17)QDZpi=e=KuMd9Puw>HvIjV|NlAW<t`&*`QJn7OAq~@
zLH~Tdf0qCISLGkmMnYmuAL)NQ&R^H(&yA$M|8@V*ckSig`yWr2;jjAh{Qr9H{{c4$
BdDZ{`

literal 0
HcmV?d00001

diff --git a/test/models/data/transolver_output.pth b/test/models/data/transolver_output.pth
deleted file mode 100644
index 1cbd941dbfa32bb5e619f5973144e8a315dcaa53..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 116814
zcma&Nd00(<^fp{XNh(UokW`dRNm6I8n<PnSAPPyL6roa~FDVU5RE9_;2^mTfQfIG7
zLWW345i(`YluXHce$R8gf4$FhUGF~EzRqy%z4rO6&u6c7ulruxd0J-$IXNXIx&P1B
zS58STB-nf9>Qw<@KEVrDg@y!%hKyR`9pXJQaCv~8je^`f*%sMpLy+7wxy2!?f*1RZ
zTo@YSAFx_^VX)7lQ2&4>3qyQlrw1<#GgA(dH#T-xlwD-C>`3pWK0%%S_bJMD0js=2
zX8rFUm4iCF%gc^24N`FLEPG!Z7#h@N#>>Evpsp^{<cy~&Oq(rx$%h0f8qcum|34r9
z=ZjLp;>D{10(=&S_^(>ITG=UhiBGW45?lYpA<9ABhRzI94hiZ$Ge~9iNatyaN~7Jr
z{mz$}{-53X|2p=5a^3$AV_&o`#Ant2ShurL{9l8YjX%IYBqYG+e=Nu1g-%zQ)&JT5
z@0R<>ng1Uw|BqKkncCTy{=a7b$F={zw%~_WLRk4kq3@AtLi8k8K~ZmkFn8WUp=-bf
zA=x}Z$ShA5`WwXvou+sTJCoA|$HF2ZO)gD1XuC*w*)mCxD<3LIY*q`lmO}->L_@HD
z#a1zz9#ZT2YDszJw^F$=&r0RuJ4v&5FP2#M7+o3|swg(0Lh9^NRJv!Ki+H}9m00*<
zoOJk)MrqY<doeRbOFS_rUz|Qnof`^@#m4&1+<a>~?>J5D`=1RLx=iO-!#SMOyp;>5
z$Mdy<b6nG!&;4H?;W3L(vtLXeFAgc?<n-HY)q069|9ix7V;*x|>r-|KdC3Wf70W7O
zK60Am4M%Nn;$AD7xR2o*&Y$*}RV0tNLZy!X*k0!rmup<H=mz)PQqE`pRr0OUhaBv9
zn{B>5W-XuBoQsc~WZc4@quz3^<r`l5{ymrbe&+1xAKaX{UC7*!F4Wg12#yU~g_={T
z!pfL4!lq4`Lh7z_LYU@l!MJ*#aLn(t;C!)8_?C4_kg633v8wxpBZ29{MeT#a*VkIY
zzN&1Y{=^93OvG+MXSYOXe{n`qE4?R~-X*8he4v#iu-U(~@v)M$|C0+68r&#(^Q=Tt
zzG#e<`GK<%YyS*sVQP(4_Sj@AxxTY3Q>@*kA9Ux7LtZTtKX}E7>0>oGYyNvtYlA*d
z`#O|&#hbIE)n5MU>&LNYqS(4Jh98el<HC{0SbJA4-!?qL8$O(8QLmN{O@6^yhL`!l
z)MkF6{fIxCb@29x2IhB7{K57qe;#w6n{HO~k<&ML{l;4?d0E8)9@YGJTp8ClT;h|)
z*LjHl1#W76$eEXJ^H+`AoP6;)7YwTBQPMh|x#$sF7JX#jyRUeQP75a`f8k5Yc|zTg
zLcuBbs<18nl;CrzTxiG^g?8N<!O`ojFlKd=uu$!;(BYy*D`&hGF6sOd`uSH1CySbd
zy$`Pms*#m~_umIX$TVwV!)r_7O@xsU_-31ME_R(@Vv#Jk&LYXz&YLB_XO9)+imU`a
zaH7;<kfTtt)=DBr`z?N+e<scByixeP(^$w3(l7Pu>n_pDB<c8tE5)kc9%8`#Phx3{
zJeOuE@}au1Jnz*szSlB@wVd>L#J+L-b;1gk?>NNAuSBx_*bI(1vxU2Orm)+cY_{%o
zfL-QEnIo7(Ce-uIvrl+lpC-Q3<tjhgc$qU(8~9sv3s2sCo-3Q{xn=lsUShy}H25CJ
zY$#{-Q*XG(@N;~4^c`0By}_t@$W8Va+283gZ>+B6kfd6Ene&c+W<Oy&h0oll=SQx)
z^^xx>+!5AyD3PSKRnV<$6~^9Fqbpv`g89|Hv@FPgI&U0Bb1I$*HKRt-m90a_da@y{
z`>z*8rQQ%Mk3AQfj`gLXr?hEXRfBLhdm!odQlYNzdQg;Jg<!ckQ5bAKRB*lXQn*n4
zO?X(UCA2k`3w=*M6E=9=5MJJTFJv#cX;rVPDg0D=CfROVA)5A2m(7=*`0@h@>H%j(
zzlU>0&v0$Ac;absq_Z*GjNK+0l}zX4dhkk(9;`6ii!(Z9^Xakk*>3nQHt!S4Bdjx6
zeeQm?QWg2%klk!4^Hclue9nIJkV`)_@~5+vtep3W50ARU7gyA?uX{buPrbw8*0sFb
zs(`m_J<1{Uf-7>1IH>v>$GthtL4h~9@y`Q(H?xc@SC(;@LJiNGUC(1WKV+}TwY;+N
zG3%)|a&F-ZR$kYG;ywCNo~{7}+<hi2@#;eV?A{CChIc2w@4B@7n=z$F4574jI;5lA
zm%5$RCd;XVDe9pHU0G^HBYx;ms8VlA$RA11XPeL;`X{U(+?yPa4x*wOdD0v(fc{?i
zD1?}&3mzHSg3ri9Le09Nl;_-!-p!SxVx3O(SJ0p=V>7`q_p}hPMMIo7JyEDH>n$;*
zlUB;M+eOovJ%oxg%S9EBzT(~$PsCz3W9}qL6Ac7&HZ@h?ZpU5t_xiDX>fUHRbag)K
z+}_G77bbDW<0#%0lEB}tUgRFT4|B`FG_DRWWu0*{U!|Vpop%qiN^3SR3Mu6D?Uy)=
z@3LW1HdiVhV#NVR*wV3(e@vJ8%IqAM4Y|VhPww#6S&jT8tDFsMZZXBSaQlH;UetP(
zPyTn4TmA2HvG9UxLkE%tk0fL0lBu>bz4e$tXU<QjEHi;B2YS)`GJ%vF7SWmR0>%Gl
zMz=;dP*(9g`q9>pI_8;BUtt#gylz6_`eyVePMykn`_Su7QsI!j8P(hxLFbo`q~CiV
z3Z35G6tb^;6`Us<(7T(%X@{>pEjpz^E3!ulss~I-`Nu8E>RBg*o+_&Jv-!Q1`JOmo
zfax~jmt3v*HP}%2aH>!u_xGdZjm<IXcztX3)7RiDy4^T|Q^kuK>sgDtv+f9gUNn(7
z<?nRPA3lcn1We`H-{*M2k0WgEev<dcMX^c6GoH&uJo8*K-};csHk)^{%Ia%8bHQ#t
zbMzQjw4P#vnK$^%j}yFR)=tjoILj_mvia1sN*+<3%?j`9`QES+Uhj8}!z@p8isb`d
z@cj|L7T<9HxHsG~vzb3;j-nq%v*>tDZ(9AxjPzQ}sjR?|0%N*U-|y>bs+KburCE{L
z_%*b2m_M~AjU#*2@sw@Toqq4xNONKx$os`4(mHHH`gMVnx4@9>W@wOe?08aFu%or(
zR#D{R|0w&23T2hl3$Ew9XzK2Dbj-L#sM*nr9xfe5%8T@=-}?TvbN)1%9%MmP6`g5f
zOti3dQk!IbLb))rM;9R`;I{B`<W8}E%r$YBle@^;zoph1&C<HDBY93mvS=~aignw1
z@bq(c#Ve^stkyf62lmb9Fu#$!R_Pf388(YuuBP+KfLPWrkK;eTlezubNnTTUoHso^
zz=2UQY}>bnpAJ0BM<X)XIPVm9a0Um<*5IA72YE_$E@#SL=POR7{Nc_W-fhC{vEdra
z={B*-e=WSPcO}2ycZFL{TxSo<N#t?bkIH@QNxS=W`q?d(qE&3^L;4mfd$WkzuQ}21
z27BsMx0HNqQYiD-EV{XIH(i_-N$PtdC^c&d84pgOx3+O)FmVqJ%uuD6>b+EhXsUPI
zOevL~R5Nc4-AtN6#Z!D}c!D*3niWHTVjG1WAGT2ebs@(G`ZT^<XHq^O&@|J7LfR-*
zijK0Q(`P18$JSQis+$4nDoztzcOH<`mo^J>Yc(Z@Dl-Mkf@ZNQca+3f*<0M_VagMW
zz1d`SH|{vKLsZq5=jZ{)*~4i(-|M2vxyi;nart(>f5MJien#@ujoaBg=m0kz@!+Fl
z(pfS$j-`>zgB>@rn$9X#ED2%B>vjCKZa??yQOIwC68O#EZ9H}OasFEUmMN-=zudjR
z%Q9~Bmb!~v)>6ao;;u3+tmdpA>#5#tAzl5lh`dU}DIznP3KbJ5^?VR53AUv!E%9{v
zwLA6qTub*J?x$W}W69`PHno&(C9T2@bVRNn8BI#3CwtTB!9h>@T$f5^zR^@Y=O|6D
zK1kmu?xTi*3Dh+wo8*kP(Xod|NzTxk29Hssensa5zd<{w^N53FCRoz&&Y9%g6h|f(
zjcM6DM~dzqL8fm4DQEl~(&#gXRE7=|#%$<DjW?=<{)_tw*|~MnB!&IL?tc*M?spcS
zY>E*pKTH!pUx1jDY9`(^>%;R_7Kp!F0{GIF92T{`*tE5r6J74Hw?QE%I$QDnWv93$
zgSgM(G`94c%>jSYcv!>={?vYm+p7IIC?uI%zHZ}i#VGEy>>S@4bd<|`9pWx(O)PzQ
zmKC#0Sk3GTJANzTkS@ji#OW$Gd(EM{-NLB2&VeFBL+QW9om6(@Fs*tVLx=Kn>6CRU
zb#FgKM;o*0T={ky>$HgqJ(>Dur_=f`>7;ielO$!kX~%#B^8dDrX8K*Bdb&Udbz^9U
zY@I%uf1c`>%I0e1Niy9NN=s@3>746aa`a|8E}NG;X*oGvFQD?&UF16>lPYNpt)FR0
z<trSi+v{Capz2KdX||+47es$2(xs#$V#>_<rH)tCNu4)J^G(i4B7U2RT@RHA)&14D
z{2z(VyhywmyIV42=mzfGtt&6z8PD_gkLJsvraYv*6L0Qh%B!{2_>Yk~r#bE6nVt4<
zSnUxmoxG2A^QUl;rxQ=Okij+%bv!QM1o!`w$9?nnv%X6nmv)@w0+XX0QQW{)-7>lV
z&U{YZnZXlAxAM0UH`sCR9p3Qy1X=s#Qms=m#jlE^PX2kcbwn)f+7U(Cca9UkKTaPK
zcG1xufn=6(fWEt5BfFgk$m7*@`Zgz({&}a-<Rw|ubn*m!aGy$FLidtx`$PIZu!Q8w
z?$J#F^ew%PKHiR`YpJK`P{30fIqNjd>HCN_wP%vz<Vp$&z9u}%SV(<_X4Cug1*DMB
zBCFSi(VgmJv}3LoZBmL8?zR}x;uk*DH+;E}J3EHr&dw!L)g{lLx_tAFHU&0%NtMjH
zkt9c(O$v;KH<MEY-Ow-Mg0;nbPFldNQ}nsFhZbLXP$gDX9uog;GUl+<n_^6?z-f`&
zxUSnyzNB`H=XkmE)&tq>6m*#7{;g-9Cv|*$StQSKdcdi^+gKrWBab{^$s_A>Ims@Y
zOY-)xs>>msXpqmYCRIFpTpr*5mQLXlPSJCxb5yXukjm$srS!am)Mk}Oii0CbzkUZP
zZWqaMYAIEmZJ?iV7fAj4P8#8Kg0?pwp(O_|(S)cR${AEenr8y(@Z9&bY``;Ws;Z~1
zU%pY_^}A`x$^tr|H-_qFeV|d}e^U7DHhO5|L8Y~aD0+i0oymPk$F6?(U-!b*U+XC3
z!%{jUeLyqY(r8cC7V=%6N8Z(ibaY5(THUmbX4U-><c8UwtGlj98WG{t&%%SZ=eZL~
ze+&EOY!GKDyb*^_$`uNqpBA2f^<(EQb`rHt`fTP~Ch9q#6LUQ;iHi<pi(#26>~``x
zm%2vrf{bu}eQ^N`Pc*sf)9rlE@EU8*p3JS6)A;hX39J$u%X2e}IibdnUC-@e)BBsb
zap)e-4QS;LOP_Q7u9v(v^f51a-^6RO_mlaGQ#8*dgT~z7NsYM$q~ugYMGa~6XjeAv
zx5=Yj*Y?raV`r&Z5<u;`?<m=pY1i_HB;=IRlg;la)9E&?zqW(?4oNA0<$ZeO`HK#O
zAEoiqr&PA&Gu6&~N~T}y$?v&5W{&?vN-cRbEc+_W^tnX+za65tky#X&W=G)@kJ8Yw
zGib@GBy!G*r8y0Use9j8>b!b~Q2wfh4z52-jc$9WnB_<-F-LgzXA@Oh3&Ng(6+-%w
zcv^LSgxF>_m-^q8<w<1^N!8viEZq7iI=3{J{#3IQ?aY+K>(#!J`x68{Xt9~QdV6xm
zyE{BGcCYxz(Sm1mOX5(=du+A$src1t9|wkX;<-KY`LwLwl~uVcm+j8O&s6Z1;c4u;
z(Tk@(3FQ9PsjRg+fvs%Hx#-9_c3$|3oj2rA&6AVl>VK2&A8Mq8$z@dJnMnouPiY{^
zDD?hm>XCe#%AVxW;f7)|KJt__&$p1~-5}aD;xl>WR8sz>$0T`mheBs4;XkwUv|-Cz
zivQeB9Y2#O)KC={f3;Jy<8O+*l0&7Z3MlK%N3waLfu)KK^ySq@Qr>ZcW{;6lxR)Zt
zdx|jJf0^Fh>OoF-t!PWdY0B3wqaaQaZ9J}#VcdQy3(}<ts>xK)?*Pqi*hV|<gwQb=
zzE-+mOT9{Wifi6)BD<%<S#^1`AZ@o04RVWxUTYM{RsNZ{#>7*)DI%J$)%WIal@8*B
z8V6qAJziYmzk~xtCtj4bN^F_XpW75PxoEO2+dn+UzpcQ1G=}k(;&|SfJ(-{XI>z^p
zEabrClRV{TC_mYIhATcNady|oTwqwn$My0_JMa>{x_yJ1DvN2t<->I4dLsobIzXx2
zPm$#HbJ9}Dq}q46)NNTCMXNrc@HIE-)9IgNwon-@xjD4bqJ=gNQo+$n@2TG*ZP1%T
zbWPG78@rTHsaYw7s%Fp#8zlq;{G;82I>CPa7pnZ#7rC)rU}K{Jlg55HsGdxNm-fJt
zPx*8v?+*FImr>lQdK$TCFCBQ4Kz~nHP@kC($!@?4YMuF!<~LoZK_AOWb%#H>B^Z+D
z(+l)&MLx~z_fas5SxhhN<-``x7_!(BAZX?#i^1KAN~2GT*WW5pMo=EN2DL~ID2@>d
z6V!#xA9sl<6T*4P(mYN*qs%pK=fr)*Nn*j3WBeg+6sN6G<K40x<{Y@5x1>*E?~Yh*
zvpvK9y*9Fmo|KP|2;{ek|FKWaXMS;Q2OqV+NiR>g(QmI)G{*E2o!fSgE*^P73jJy*
z%;hze__Wi<Yq{jKyphuKTIgwT2{kEwBS@apnoCD$>_$2GRA!UI(LNBK41(wSzHscV
ziq9F1)V9tDGn%XDy^|Uy7N4Q^9o-Rnst?W`(17{$cKVzAhra9U!M0`)n)(_ce_<c^
zUTCI6N)`0dO9Ofl1292T77qs13e)0Luy4Ksel)jG!NCrC;J1m6etJh?s|-k97Gt-+
zkl~gWd#Ks#8u@A{lkv<*vCF1(8otm>aQ+!f_0NA=>6PX1pcV;9PM#4vSy^-QwFcqD
zNqf<y;ux#+{V4g|@3C~rjwb1&BfEL{@t$H>zCIrvw_V)rUN2txn#519pX0wfhw$gL
zXg)M2i;r$2*44en<9DXB#f8WGIirv}-Q2?)G+xrW{CgDJ_?JG8P=Tn~NI5!>NoVSH
zYMl6hEKQo}mfc^P{HTi7Mr*(|T@$<Aba0_p1NHvd8`=B1fG^&p+Kt_@`mhe3-0y?o
zkGGPoM<28dHbd)*SG3{30J`Gwo`TEO;r1z$?AKZ1{Jmb-t~Lq<)f)IU<_P&;9f060
zWk{Ru(cbNi)PA=Y5?fp7@$CJy<9i>hemxL(^`-QxhbBI@X;IS}4TKu*qS<Z_$WXtD
zMm*dnT<)1p`(GZU-3|@ZzrQuTY)T-<KMHK$C5^htucf|A)j4UojTpNlmmgj4Lko`u
zi#gF*T+zKrQY`Nxu3vG1TkU;$_xpol$mTwL|BF0pHivMG|0}+}vWnXhx3PnPls|;V
z@}5r<S?ORN+bE~;i-mVNpnMz4k3Y^KYg<`HI#Wq(R|J*Sk!M^bB|UB;@5%=>A+w$O
zJ-top@x2h!xr2Hw?E?=JO-z>`fMFU&2$=kb*4!+mQd4<E=e1L}ZTi@9>lHow+8gcF
zozdyNGS-<i(Z}jb<maP`ZufPt^vxv9Zfha61wA1D;2lj3JwT_Qjlp!4?l3DGi$xdY
z(7tmowQVRT*TUY|^Y|~7PwD~dHZ|zgyr+ttBQSc*9dhoZKrf5yY002W(%UPK>$9)Y
zq$SfxvQ`Oq#de|C?EvX%mXXr@T-tA5M76U5_|C0-s?*FDO@`XhPp=h%oYxm|b=NR)
zcb7wa{U1bky-5<2F5S6U^#z-xmh+r>J-NK1tu*scG+#gWNX(9l<(gTkJpNNUJ8c=k
z%D2-v#cC>FO!DANyYswdvy=lfb6M})UVaka6%FwZ=xNC(s_?r{VrgfjZ~a4$r?t@A
zn+>F_{)7U?9w4Laept8eF|A*tgzb`#6c;b^<1H;LG^r$qe#$u4rVhDVI{19W9K(O~
z!<?VC@ENF#<>DPuU*84AJ^s+?xFP80_kf~z8RPbov2bu2i6u_jaJ;C87P(Pyb5z9s
zVQI7~rx$cqn_=p8Z7g0T2kTxVP-0O{8;|uuy1zQ?ULPWVH&cANvqt#&X+PP`QoxO9
zB74Q|XkD63Qv*L!zS3mcaAY#+)_0|M=I6<@ag)@e#9K0=`K=K1YYSCbDY4eT&QurF
zNB9{wLQIqYBDC%lxOVhNA!I^>)cnaANsZB1zBSx{v!+esAib4*-nD{vDats5g^j#%
z-C{oFoyLmNvD`2A8i%h)<H#{}to*KkgKzXiQvVm^o~#S~IXbwP{Fio?zo)>@FUaED
zZ`yrP3GOPgSZbt#-RWvryL1TFb^1kG&e||h(uTvzUNBv2iO-w#5%jkkvV3&#X7*sT
z$~i&7ZU(+-x*&MsKs+1P8FTFcX(t!tiodC%w=2$_H-pavBiweFjE1X|@%ZdB%HMf}
zvNkxP(pU#-8g}@oo=(@y{*m-*6)l?VfE!IiV5vI{2E8;f&mf&LHfdt$cRiHW-=%3L
z`s5r}Pe&hKrKW=q1>bYkWb)u4S$Qjx>&H+!QZ<EkB}R+MH=hgd+N#9k<F5!|SB$9B
zC1o;LG?DvQg-Cwx2ohpGSo6rm2|RztGM1{z@mAMyY$4+ekA6AH<;O1b*sH~ywPFK5
zaM5F>12fsY+iqSTbb))#T+XdE|F~D{KwJ-gO;a825ShHEhgs^VjeJ7~PItnf7&%<q
z@|t$5b%#rD4ctuZjI)mYpk^dNN|*r#czmI^ntjk&Qw66+f2UsGztDqsdN|u_BHULi
zV_wy4Tt47|W|<F`&Kr&)q69Bb48rV!8E}0$2TE=>uxT0rlT|VFdd*Y>eze7N=S4Uf
z@tO=I+bCnqEPR@;h`Mdn6!x~BhGu2cwAf^ta;|~uRSXd5=ShY0CZlMhE;b7iOnR?}
zin;5A<GF2AsMIW^R}7(wxLxF<x>6i;qJe$|m{P)ASBei9AtGxc@9Td<4BWDr$s<j2
z?opkPT?R7kvP&v`qAmWkTPvZe&C=Z~4shD{L{`85Ub^yuDsRjg&oN0`InA>ZTg^Ge
zEBDOhmmSAA$V$HK#@<#g54y+SeyE{l)Bv;@%VTrYHL~(*p^MeuDRH|tTnbE)|EUM2
z|CC3)pBZu%H`3sVy;0s>0WNOF=#bOKy(E2fjL^aQz`+=lqKzAEE-;#BijrT(n0<I5
z7M-|2N)2<6x5gKCJ$&#(mbZpnv_;AKWti3>%R~LNvH7_pZhTmbb>sWPW_uS@)e@vD
z-7vxcus+iTYps6K_c4oMpXi0n;hkYMUmm^hc#wK=Z)gm6#lbV_!rtMFWHD+8PEON8
z$*a!z8nd0$EAG+>^E?W><WE`4<7lIAF*)D6C^f%UOLG5RqQNl|p`>@TwD&`KN?ROC
zCcg$y{#aGvR`wv_V{M*T;v68HAFj)}(UJUZP!wy-bm6x%J^A;gcyV&PJ70H?X5Y8#
zcwUSf`<*Oh6U}D6x-XZ{R}6>V$*zbz@QozT_2B!fD<-;YBDUiNb#qe0qZw^<a)BDM
zo%FGxYdv)vI~g<e%IKV6fNtYuIAGZ**uH*4E2E5XFGmUEC-lXWQx>pzXoUB7<*@AG
zBy{hljCVsvV2z_K`fN7E{e_<Je5!+@4gH|#?TEFqxTz4n1hJpG!L7$*@|@QVAtRJA
zv_m!zYu)i~;C3o;@<FzzCtlyTqmn!Gp@>-+7&-*Dlm3y{p8=Q$W9W=Bh3BDexM}yB
z-pR1^=ews!$!#vx-<>OV9dwIosg&%>?}<yMms7u;Dg0H23wr#yEqKn=6IMA@3;LYI
z3X#28|9F77c0oJ`PTeATmOvc2KvPs!Nt0;?pZF%*h#!9S;@hQ8+<tyL>+5#qn_kU)
z>+3ak{glee38%P7w$_X`>fnds2oy_tK<~>hYUpl?!9zR9twMqvSsCr-;f!{v4xA=U
zMp=Oac8nd3v5JQ&Xx9{(B5R2eDW*8?Md<W(ICf7qLGdgrI9mildQcrNY?ooo4mB7b
zu)xgGi_tG281v04>2uWtOk81xwJL)kfBp%@n@)sRVsHF?ya+QIr($K^SL(Q(P51Uo
z>B#McSf_r9%#D82^ZSozplE<Kwe~PuVGpleMbyL75=q)Kpf$-Ep97}Q9G5B3u<nAJ
zaz12R^qjgH48U>c<HFX*t4Nw+ELabeBa_!1LR#Pfp0jJ2j6WMCnAXjbxOXv-bo$vz
z6fSh7qXujFZ9yX6@;xZp8=ezq=uYR!O9rsRl*^*m<LTm!=FL3*JF%;BC_jxi<P^CK
zPOn+cA$`wt_qJY$Jv$huLPw)=xja4|lE<A*ipUz?9kwH%)3b%6@W)~hhG^>Hh_WtB
zvu2?4qy<o@(Sov04OP|o<3s#(<V6g{=n5k!RaijcI~{&%A-JXMj>qwxu;_{vuGde)
zgyUAYB$%OSK{S5(s~}{Y8x~FNftbVy7__a0LVrsnY;lE)_GMZ>Xd_5}HvYU1#VT!O
z?7BG!M^^NOv$+qdO}5au)$i!GTW_2{nnL-BjWqe^8X6k>g^Hh*P|lOR6uI~}sSc6D
z&#h{r>WELY`+Gb2H_ar!q%YJyW-;9k(I%z5Tat*t!K62LHQfm~O2Lo3MD+zk?jd<>
z?e@j$xsi+q9B##Hy%NP<<EOIbkR9y$VJIhTTgExZQ`v95$SWpH<+gk$4r$BaV#7jK
z7gG6thXWQiOYr2e8Q$piM$h>sFx&^ogdjNI8iI&iPnePxmL=HZ#zrm3$t^^WawSBS
zE`g5Q6x1b8#;&nGxZ0O6{)iiPUta?g(E>;Gd~scCExzVBLF-a9Y6j23SCjeZa?lOk
z^EzRm4D)!&xPUIY2`H)UhP)_3p~E2LD9K{R9oe{#Tf*hIBIYdFjIf8tsE_ClpOxK^
z-@XE)S_5c*YZ#_~Qbh1oV~VmGj{Oe%sN07#G@yqXJT>+R`&MnD?{igA5cix+!Zl#=
zcDoRv6i>x+1F*HA4;df0Ob0V7>1fU?YMK;fm4DAc%+~Ohyn8f_`nHt{otK*Os)xhr
zrRdMrs%BERZr%8nayRZ|dr2HUsZ&}0n4^4MlFQGvwsW~>GD{6EaE(?4&wa-{WrG<`
z2kN3@lm&|N=3}M3I$rl0infl4a0u{0%<~C=mm5AmS&ZH-x)^$01C_G1McqbXOUV>u
z+fKrk#z81}wH()q%+O=-Y8Y2VL35G|Mun(hREq_U|CGT0L?SBh>4KK(VcogGFn^m#
zfhmLWWN9E4XPZDRG8pxdf%x`!3g*VlLA&HNabY2u?Nh_~8>8{`#7JaxSAu+h1Mn|H
z%yij==ms6wFATtZWo7K2N_a4&HxAv@q(<FyLW++msxP0W`i~Q_Z(<sCRhufbh78A#
zBW>c6Woc~hZ7m)j?JCIaxIXTu{ws0U*%Z3+Jd%nhZWsO96D6~b4I}Dr$UFM6P@X+n
zawIpBO<Xis;p`}(;!Z3_S4nvK;}zm~-2wdXZ@ervk77&XhrBzvnY+H%L{6V6n09L<
zvbBtGF|02x6j@``Utm!Pu=3+1Jldd*3=084-}>Xjl4-Dt>Wj1f4w#{`7OR`>@Mu6a
zRR;9O{ZY;sP#llz%M~!e&j_#Tw?gxcGGh0Qhu+rNun!GJQ(G_$o#b)eKM7gRt5DXc
ziIskY?t7E5WAiHH?VOF0rE_s!Uja3ZQ{a#khaZKTky;suyi_lID{m)@2gA@FXAYeK
znh35tN&~xVW5x7ul=6NYP7X3fuh$o7$i*4->3S+zb?bvATkNn?w+m{1C5hIJx+oc<
zf=!3(gq=AG*ckGQ5?&pqkSWfhbk{nn4?jrKs8Otv{Z7nwQV?2pT2S`JOQnfsBiU3(
zTk5|(gsXKd*goL|KWQ%J71v@p%~pl?=%#SUQw_E?7{%)Lj!=)aLBDr>5Tfr41tmA+
zY%;?4VZ$)`$s+VVKM2Ni$727}02l|jLDFX;F6#(bx333&uCzhXy=XLF3d613f#CPH
zFbf%nkh4)pbXLZ1pI8_Uh{T|TRD{cWAVtQJgxLGz&bTBj$n?Rlh!|*(RK?KqNl5lF
zL})}HT78$JZM-+eS?AM=jA*nkcu6<acEaj{5j1>#pz?ejGL@FVKmRZNcrpQdwFB`r
z><874j>L5bQ_{#A0?ik|qlsEb>}^i!IWjy`XM@f@N%T;0AntW-q$NGx(cR^zg_Ell
zuwYU&N&5t{b;3Gw`R2mE-lmJ1mji{OuH$%+`5KmQvEa{Mb)wwho0h64V#F60nmJ<j
zc5WR!R$O~+GwW9$WR0^a{Me{V*~_8^o*;(t(ly4IP-lW^)0}Y0xIY}a&ccRv2kdJw
z#+dwNu=SXY)9qt0;h-`WsQV+nC=e<SLZFep6)&#Z!8tDpEALH(<)j^m{BIOuQw`z#
zDgu9sV&JG~i-a?Y$T@5+`|qnU_untFP@j)@vjH%kwFdh-MxZXh4yU{Cz=jSJIMu~N
znm!EaOQzyXsen(HmqBMzSA6vB373D%u=-3SUXDt}x?_b@+N^}U;riGgWQg8Rd%|;r
z8>+9kB1PK?1J^p>!I|E;m(WfwW+`-hy$V`J4aDw@zw{d?=t@6BB-tj>=W#nEE*@v;
zW!-;*!ZeZG-)?2gG)=y;#)Z?aoZ%A`!|sniOW)mBW&0Kd_R7B^wprA3T>4|tRO=|G
z-;NjCGOvhpjQlv}sVI&-m%)1rGq_*GG<;m9h7W~(5ggtfQ&vnt?^**GJRXE6+N0oH
zKLon6-1Mkw4bnrW<MHjaSiB|<!b)3o`e}#kzN>M&!3zz+oAAUu3Hi%o;rr1On`X{K
zVVCKc-F**=mv%y+`4)_Jm;?&V#GNJ4FkCbdjA(d!q(IMoKT4aEFnEwPVorIGZ*UCK
zhGyZwhE>=zR}KLU|6zI`9lZBi4bz&>WT+g9<=wVp?WhzS`rAP@J7(gkof7mU%TS{d
zLL1(A<LFvd>@Smt(*9a1t}@20H?CYcqL3_B$?~1;K?-{y<E(7^ac;pj@voK=O*y!q
z;-Z_xn1FE6&~^j4HYACQuG|tLG0Q4c=?1%NS4mYoLU@Hy1~+!T%I6kTu-@>k((6tc
zyx`h)zIHi~OQwwC=27xx>HnP2`G^H}?6O7L16vf2n27p^D^TWZje)a9BEiB2H}xz~
zqN|V0#&w9hVu-@SW6%R@@!4=T4u1&2+oTxmyEq4ljpJ}o*&MF77h&N(b9kuPBRy>_
zT;wgGfCJD!<bn<FeW3Af8<Ng`q1JS3yx*S<VeV#>4+z4J*F&+j-3S95cR(Yd7mi&`
z!i+QCh-;Ls@quO3aBdO47yCfYd<llXOU4*OZM>8#ByY2kxcpfe`Iq}(bJ-F+p5_XL
zuzH%KEv3HJ6H$3`Ji-*Z<5I(Ld>B7kko)|-<Vx#OihSCQTo!0>iSq_ByuV9Q6Z%n_
zKQ&VroN7n`!XWYe>|IoTVG>!mw26lO#_>JrGts^IkHmA{SvEb~!e?`3aW(so_+iRE
z{&~WR4}I!XmTv!=m);OiwS6jrqh{gr*h%PpE*vukjK@p^1yr9m!1ue>kVGot%0wS1
zPwI)a4l-V@b}bxlF2I)SY8cp%1k2iuXqdVNVKOXm<YX}VJ3Hce<v?6EbHJE|K9KmC
z;CICsq&%OG$1XOwe10=LO=8hcX&?L!*&%6uXPmyUAIo3vfUO(wO{o_)i~G^9_Xo-_
zO2Fmntq6@v#J^GhXuHE|EK?hUdp63jb{Yr^zs^|iX^&4|tP%N{kZqntxB3dScDpxz
z<f@~|wUeNn`hsFwZ%g0w8HS6`?~q#K3wpk+mWBn$VfD^mG;h*gUZ2#3mS_xO^I4Cn
zXySP?{#7JdzB-lCqI&Wbzolf|e@*Gs$A|fiOtb2*P{2L<MR14tB|c-^wXBnKGHbd2
z;IoB^Y_;kZYZp4g{{qk|8sk;_1{{=Mi$|w|F+FE4GFzjtj4ZLkU_NFan1h!Zj&QMY
z#lz&yDEbhKIrpc)W6^3%Yzl@+Ogh|RhvAlVG@fKHhPG`Q)ZZs#to160Q={=}nG?o^
zyW-E{1sIUH1B2J>#L=1%wAaO;B4Hqo+z!L-KXDlIG5~W&PeJOn(a2bL8136L@V-0|
zBdU{7`P~QO%%`D3HI??@6XnM(fb_*UBqj~TPIX|zb_1kzdM3>H;DN6mKj~+`YI-lz
zb`@7Bpt8b2yfZ(UmZT-}!xbt}c~L=^!n#tQf7>L~{XU<4e}US6B(R^4hS=zLPb?qx
zM!G*Ii^pD@K$*?PRCrHb`n%|$nBH5Tm(IJvRklgO8<p|m*$!_W^ym&JM_=Rv1<Zac
zcJL>aQ0OMDLW&rH0iR}LX2~iXE|T#VhEe!W0@NCAMnD%rz4CnYk=;j%E;3wtQyunG
zC&N(YH(cF_`!(@cwa*57lZ`O_#UXeePDQc9MC_^CgR$D-FxTD$<(hN|DFW^W%dW*o
zF!}9L_$9uk#cqLkCA-kzZGGS<>lMFmxCx`zO0c{=4wKz?;YsJ4^xh^J`})>W+lk)z
z{dosE4b-Mhqpndzl>)lVAA#L|gJ@#K7uxX89^0<U>fQE*D7M{ABi6*pdcBKj$WcdX
zzN>_xY7L^>;}fK=R3UxRc7Tp{{G;%BTZDk9U4rKNuITk)Dz)t2Bfi~KMjaYiJpcDv
z4%<JPv>N4kM)_R+^L_w3U(XX{4PbnxG=)nR-{gd*OYGTU%@$Fw*j4OV*0&e)!Sg-{
zo#cT=cY7#}o+hjNf)M4k5nW|D>$D6fIh>sXTl-nKZEl5^>dx?vTaIorJmh+58QNs`
zTy!r1QRTk)rey)?u~gVb?7)fPG1z&<3+INWVOnw`4kV66w~!c=_7BCh^)gHwu8%8&
zEfK$E6=qbrL8b5na-VwRpMMG_PCS8!{}NDZvKsyVhC}*&Cj1xdM|h{rkUZ2uQmZ|t
z9^8de`8fD~--PS3HJ?An2ZFIXvGqq<cyR!#3VTBD!#A3#`HMnI=A&P!GqRU;(EI5R
zX?}CM(A6eJ_-Ef8FE2kKRkdJ>Kdp}EFB{0>S-Uv>-EFFwutxf*t0uo0FhErKyn~;=
ze;^dj+9kSe)#ceCAsnUV$h)P-c;0||u3S|o9$2Wr*`@j1M@yf3O<&K81};Ux^R>8k
zWDJU~2BWpF6ZD?P;7od7G|6hq?q4IJuA~d24O?M6DGc8(uD}m<4{TgM7vm%b@Uq?y
z``erGVWkfu)q*hKMFfujXO1t0ahMY3gIUK`q5rFGcxpQbtF|A)%(fhy`jCu*qSdfV
zodkth8SuRqflj?F;WXJ8Z<A$Ok`AEs$QSdIPSAA|S9oe~gNBALrbmV%MRhkk*N#J|
z#bX+;yo+#SGJ?wtF*L26%J#02aTXh}@y!BgX*px`HeJ|en4xrMPdJ`lPsT-hH0hQE
z-9H78;g<QL>STc?jM+=sK9@@)`8v5*-6y$k83KQ56@S;s`q-{okn%?z=6E{}uT(Dk
zbw-u4qXw~@M>+2vIhAXESFmlvQJ$6K!I3YIaptxb?$>sZ8<uRvsa6A6C<jANc|5#7
ztVSo>L<~tMI5<QC4>!Yckq@?Y-i8r9BrusY8+Q&(f<<+R414?H=$CC+`7{oH^|O#{
zl#1ZY#kgOv5Svp{F*aZ~9@Z@fWSWNCNJE@WvqGxzRumPdB53PKtejbh&2x95efm<&
zH?hKE-JP)9I0uKaPoSl`iLxDRFv)2Z9$d}A)N}K2)btSI#ThtW?hl7m6+*<D-I%^%
zJeJ3*V3s^^C(0f3PFM;HGkZb7!4cET)o?bmhB_vi2=&d;)T^JbSQR-D_7O1xrGKV%
zr~PT3odMc}R#JEVB*n`%%3J42J@zNk;EF?R(^=MUa(1s&K5-0p@(SS;18uQAI9`lX
z$rB>K_LHbjN|0`;dCGQPvKlMrD?85CDytY|!@AD@K~rrmgl#fT<YXiqr@JD3f-A0^
z0ABmf#$FE#=yvhI)VJPn$X}23zY-9v;)4s#6EXa%4#IdDoFgC6!9Vj58@K~A-`d08
zD-90{J+UAx2;t#jm|_}(8r3ma^GTMwgiVmI-j5Q66-ZZ$Ms|fWDu?O7RdF=pZw|wp
zPI>rq&I;`-V)4P=3Pv7oSedDWahjQUC1(cx-@|b&M;<F;bWl|=8uB0a6SfaTFV80O
z_&E?IR`IB-EuogbmbiapBr3X#<osEb9%?vEdM;VwFBz)SJJd>biE^m19Ev2zRn#}|
zkx(+@7}@mNL7T>=2~V%CrtF{^!7l2U_`}YR?>y7xq*uyRtg(~3-OOdBZG*(jb6M;*
zXfcQEOXAGY``J3|d}&dh8yh~}!85i_#noORs5n0gIXOO%9=FE%{?1qwDnZ&MO{96v
zK##1^FlmlP<fdelMkzwFE*aYn?Zx~PvFNNc9t}HO@Gg5Y_MGrW&fOS{@eapaSzi9{
z@?gXtT!}Y7*THG=OeDLGN8+PUOqh|2v^FOgbOgfwZ4z1ncSC+i1n%fB!1hcDeos|I
zVO9vr@5I44J`tK;V{mHz0?giMj%xX@H2rvIxQ$Q4+-paKUC-ars_-eemNgf_Izwr4
zLLh<>O24-{kzt7+)%Xgi?686UX`org7tL>&1jnWE;?fRlk}H)8*9SCE(ATwOQl3TW
zd%Ci-TP5vuTg28{w<+sVE|upuNpDq%e6w^LrEJ<DG)>tl*<L-3b7ap*%#69qyUOnH
zrtM0cZoGwWEl%R=gVu1y_aPW!xfs6V;&JnZ58hYoLvrgZ*e~<M*@Dg3J24rrlPBV3
zmOBc5`9oiQEhxzcC!Nz#r?dgTJLh5cxmD<5p9t3n0a(-{99QOQp#P6x+>=j%*U|;3
ze!m(&EC%Dyiv##DCI}IaRM2(sR%p*&i#~@3qip3$ScTZjYKEl{TjpT>e~Ym|Wiq_Y
zJYb-Z3?2Jzs9bD-PGyHtlsF0gfvcd>%MIIBfbd2M;j`3mf2Rc$WHo(><qT4Mdz)%C
zM&hM?5MJ8!#)K{ZsA|VWin#1UcjCt5%kwdmvhSo&oN-?4UZa3;C4bWUt0k?B3KS|5
z<H)?>2N~z@lP(+gg#7meacSy1{`Gx0O+8%0_dfR!3iOM_+SV())aWq3)H%)BCiD5Q
z*+$N+ir{pon_M<99;Xtg;^%G`ROdLr+%FRs^kkammuaY9;f>>tD{*(J1GE-z!JxxY
zxRnx(-gC!6AubYK25-R8{W87t+iVp69D_ZhtT1^ApfgQY%ZE;Z!7^Js-t&_xr}*Kk
zS1Q(-TcSPL6;<Ac;rUhpb_0A6e%TG@tsl_+p<Y;36ojtjfw<@Wk$S8%z`}|{n3A&?
zLvQWFu#D*#DdSJym7J!)QHpR_r->z&XQ|_#F5aA-g(g{DeBwuf;Cj!3W_iTYkunWz
ze3(m0a!bTDb4S6&IF+)~+o)%B8f|TTBRHPF%9a_C(t^v!Nm_hTFdXk9j_tRBl=Pa!
z$c`MVg%#iFk$MINe)=L*P0JJCY>VRh(W6MNV|A%@inBOj<Th@s&k;T^?BMq`?i@LC
zja2!y0{6Y1%l0zfJb(NWT$n!#H#KDZbH{3!gvhY^G6DZOZ$Rp34~)B$jM@iL7;WQ+
zqO&_uyTk<VWw=vu$aZAMMxdsb9-O!L#MpmB5uP{{@Anko`J++T_HPKn){er1#b#K3
zel}+K`=fKg6l{Ds01kJ<ahLZYYT9-jxMq%wK_{^2z-Zi%VJCMj1LXe<!(Z=3P#ru0
zo1Ukm<aR6R83o{S?*Y(~e?q(S<1ufr1j8n2L83Sjy64&{any6#(tk7UTC4!2<v(eD
zzz4F;C>NwYzeFo*XN>tL>tiyMqww$TqKZnJAg<U$=HWNRf7}JreJiMN-Ds9QJR_P_
zN0aH}1JW%vdR(QaN=;e&r3zIZ^rGJxUe+g>Da(?rCHB(R(BW+8bcVTO4{z0J<Fx7%
zEN8V>a(_w$7Ss+x%8UhA`%S>@scVs%8;<>|>oI@jbXYErLbury&{^9Z&&>q5mUy7Y
zXBX&|Ov8#&LuCHghx+e}Avrk@+LQfZc_ALIe%es}D#M_u=_oayjF_vTDDGng{Zm1>
z^tvnTi+jQIxdL|BufShff5CEDJkI>K3$gD9BKzeK3<^1n*V~d%o1+hFRZsNYbd8oc
zsX}4J59(#I8HN=P>B{Vv^z79fl>UiC`CT<iv5kYNUM;<yql{zv4TKk$X@pE`taATG
zZ`wCYJdCuVuC`UwJLDp%nfOgepPNiI>Cc7s>Qb>T{H>6u+L<o;`jM^CW|A9wgG#TC
z<fA)Vg=g2Ekh7r?-JCLmWPyykeb~c+!BU|^{J>sU{iN#q(ztHmPTuXL$ip3S`O=T?
z?0dft`u|>nyK94xcO@HxAFYt-oRKiD^F+4SLiEjCi^p4b;phl|(D8B5)Eb9vg`3dr
zkPi&>Vvu5|1?~UrP}pe!8hrc0L^Bcz>f=zG7>xz~uCUy0kFaA-DC#*HMd~l8pHVO_
zFOGtNNhszGe?SM9Mq{0KBzi4RL#5wnOfsE-#^Qzedfx@B)@oy7_$9i0a0$93?M1sr
zHHCWLp&`R}QhkXElsbOXn5@|}POCd@)sKXlOp_VXxsLRWyWswd`*gf_It4!U7G|88
zOXgF8No)ERu_o}O(AsMO9e=i)ybmYRpblkfKevL*&yJz5LU(L&-zQjqTrBwzHip%s
zW2I-luBP#ZXN0zuy{Jg%jbQKDl{yrraO=GwF+gD;eP}z%F@5fE7OiK^><qzY={-JO
zwhgLv0!*|_(H+TfGmpWs_jWk5dM(y%4#D9ItMJxl84}lS!|LQvcv+2wYWg%tE>DGR
z^G<j)1;IJo6pIFDVRAxeIC{8Z!HD6|IJ*Wmelp$Vn<c6;x?yO%H{QfX;qm45sK3$^
z6FUf=9kM)EZ-;Rfv*1@424@9p_<jvR?Y&Z3GvyXZX1jySe$(fG?T8;X9v9}kCHvKv
zDVVR(#Z-H&_pPVR1<mwX;}w0lV=d^fRD<-9fE&dYFs+NF-M3c5pics|YI`C~*1xqS
zxhD^n^`xwejHT3JBhc)2MEE#L6=~JUl3C(F>R<YuR{AWYnWHjf&z&8lyH943Z`3W(
z;rCH7JZ2BCbU8(H&quO$*K#h~IgC$7)v(GcWsX_(f&<o<aK7<RE|0h&o^SHU>o1`w
z+~^Hkix9kS*oEn75m;dqjg-`x7#miAH?n-+esvdi!vQ`;i&6Z^7x7goxHf4RrhEK{
zN$K(^SbQ2SM`Ey|=TdCRJ&GGkdc!!=32$YbeDKfy2=#wJ@2)Jx2wIJ=mk;2rT`cyE
z{Y4M{9;d#yBk;yPOV*>&NXP!OMtYGxu8v-fo4b1AY>o@2yqZGtudb4cjuDEbrv$UZ
zpQ-3yPdxaqGi;l>qU+E``aNSDtlg#}EcY|%Pm87me{WKG;)95Ve%NAdg<D_bFyu=D
z6^vcRt*11Ze;WvAtj$TMjVP~3iTr$9q-jy7q_(df3J07GI3#KY+gr{RF8s2goXUd~
zV7*-|u}P$NyM}Q8#`A*xG<U)Hc(izFlSWzOq>)x{I*k*PT)MJX;1|&^EFP=(PDO@K
z3Qqrvg8o?tymyH}#e-xFjvNUiM@JdA=L5?<@vvzb3(a{OWMe#ncei#TtnWmWjNXI0
zWnt)SX@&VubP#@I4&E)?j75!(_$=!mZT1U<ucINdGJC_;eGBe6C`0wH4F=6Lm*qbh
zk5_w(4)jY#um6VPvRnqrk_V!Dq!#k8)=>4LsTeQm0`C*sY0r-Zs2e{KuM@sg#L8_b
zD>tTQ^=QPE70}<|OR@NM9L+3_hOYE9%^Q7HxY;+I+An({H^7J5F9p!zR0lj!9S*aR
zf2p~pmiC%<qv1tUMRz+tdKfm2W9OZuJ~{!y49hIGiP%Cf=^)*z>duD#J%zo!N7Imq
zskFX!4i)BPO8$DwV$auwY(4#h<fhqE4tu$jHGiA3b=(@M+BuO^TBf2Zb1NPiB_XWV
z9ozd^<BC%PzPU!>-}ud#azqy+{9>V~xCzO3mSbC$0Y00B!`;IWc^T>O|9>bt6Tce2
zE{b1NDos)eNm3+9lGHtWr+H9Fl1em4(xj3^gJ#XMW;AG!N+r~N4*De|l}rgqk~t((
z=J$F3gnK^EJ!kK=ziaJzPUvo4g~Gfw_-mUDuDg8*H`sgQHdce3cEJ<Px{jcC#4<b?
zV}f>8+c8D21osOT;M^z))Y@i*i-rQwFyT15UF3n?qFnU6J`+vT_0VI8y?<?VF@fb!
z?yY7$LT=*NA)1W7s~^H63wI33K7tn$!m*&;2k#wE08V9=&e+x8;6D`1{m%~Y^F73{
zU>(c}+snw^Jw>XoZ)Sdpcz__w9=^4{%zA`GQF+Bd@Y`DebZ8uSPn*FJSWiLW0Jy%X
zqDtp>(l;{(=(x`skl41Bx1p|(evofw5^hz|XHFSlHfB%rJ$5n;lX6tMilH2fZ!Maf
zjZ`9|kzN>Od&`D!w66-n!;C(5f1Qi3K5AqA)&KV<OVHqxHvVd`$8U;BC{XEy!E1Ez
zdfqX7e?I}`L^t7yHR-s*JsQh)hok>sByyHo;p3M_ao|6FJe#VF_oU>o%19Xd(`Mjw
zaTkn$AWXm837zb$KWkPZMxIZ@$}<OX^VCpG{L7DO$AdBZVhkQv62}IYuMl4G1q9Z-
zf*Ed_Sn^B)kEdj!ayrXOtS-PvR}K7JD}u%k1<>in6s(kC8MsbX(=Dh1XSSE=j_89y
z2XPd7D}_lcCoyH`IDqgBTBybW`;FDy@mx(5TwMn)+3oN`$cd_T%^>etZJMj~fWDqF
z#x&P1g7X(2GZNZ1Ji-0;jP+T<6j$zNBHw%Jw#Zd6`VBW2l>t*~yG)PrF3q7C!C$zy
zZnn~s4+~M>bvs%)u)j$ivCb<NU&3*W{>Med?P)lh?RNsN?!j)B6D_U|zy$&Zm>{wf
z-z3k$NnHo*Yl_3&ZU^vcpgcYowZyF_7U6~(QB3LFha;~)!#Af`{J?(*rB=TLb5}Wb
zmRyVrRT?p|cMINP@5^zEVwB1D#E0TdkUM=ls>TGMZfPjmor_0i-8kHHKZVO#UrfW%
zNYn`2ieCFnG2qv9Y+O|aZ9jezS0QsePlD0qV+y%7O&pbesp0UiKr);qfQ@?{XxO?o
zi2J+%G^Ey&(nu}PPS*$PbLm`PZW1VPym_371>99R9?&OviHWZ`4U2zDgUp&}lKZqC
zij$+r+^g+O*sdBF*fR)HbU9Vx{9~s7m`_!<RWZx&w9tHALfR&58I@BqbU4P7%XfkA
z;z{3Flz1D7YxCVu+}#@u>o?-alteWB5QRbCqcAX12UGj@;oFH|<Xua`nWoO@S>uSi
z`s}bi*&A)H2I845`|!cN<LJ$Q0q1E)q0x*0oL&%sR=qxW^4w9pkZpvrx*8~;HyaO%
z#<GkKt1XsqL(_Tw`0wm{m@R0I(-x_q^sgjz6;;L1mon&H)(=y59EP>l2AHKE#h&3z
zNUyfV5tGBH`^^MjYmX4w!CbtVJsYn*-HehyCxQ2a4}-#PL8HNK@NAfg$EVGJ_&qMr
z=gr4l2(bi#pfb|#VuT%)Rq(yR9{%#5Cy)7GLT!W^%c6YJRgGz++$|~K7<HeSX?y|<
za$Tvve>OKlAp>%vs)(#u7xB*xU^*9nU|wquF~+sl^xB?s+TxT*Lv}AA5>Zu5h(rq$
zYj^;wBX^+k$|TJBwg<J`j$-YNU^F`Bik*A>(X-bKSEZWcxYj28(6<L)rYfPmu{%ly
zAH<vD3TXGn3N2d0*f~=ird)}_r3LPIDNzEK|2Gvi5A48$Qt^0$owt3R{v4b&pTYso
z|L>XE*w~qbijVf=P8(_L%3Fh49v$#7!x8(^w&Dw6d)%8_iQKKhc(1k_%$9tBxbss{
zxW*ZWA}Rm`eNfp(5sXVVVu1ZWXg@t0TxPU^B>V%pt5>0J-Y>}ZIsndMw$Lqgj<j7?
zW)$W~;84{ikhtK_TSk_HVRs}g`l&+yuwKE}-X`P@e;(8KU>5cKd6yJg-C(-)ec`>b
zCyoAjiWFX02pTVXsd3FlI*=GdI7?5pJ{-M5Yg9s^<MBA<{FSA$H+-3uGrFku0YbYf
zi!ngR8~fPzDe%-L%ufKU7utnKgR)W9eK{_Bz83S<HE>^MD3**kAV=mPZtaf3bL_n~
z+LVbRS#cOs8jsueT4IIKdeqsTfkRzss8z>@|K9AtE~W&Pa&1vxX$l(3?Z!**ccRu3
zU5t(OM>+o$ShuMZe#d`;&L?{KL0uYq<rZQ?uQ(3xw#Kg&TQTy65PIYd!R{HX7J0J{
zKFbZjTft4}zGDvB9gV>|pXXuBQ*TtgR|umfUSPbu7&Zmh!ureiVd<Mw@FBGT)EZT=
zME)jZtXzz(_S3LBNSySA_Q3d|D5{Z0A!e=%l^Q+*i{;NUw!vZGA5+2ANr<64o~MzV
zZHA;oejO;7@1hx3j?gZ(WU8GYMB6ya7(=0W>cZXypC&tKm((;G5gkqiL)EAQ&j+iV
z_am%zz&&4LQ8sxM3h)l%ov29Uk|+%3Uya9PYjEbnc+3=xVc(c^{1wH{EH(sTSZ)Gp
zj@Y4Max%W1YJkO@R1_RA!tVG2d^}u)B?~>Uy=FFA6z#w(hMmxpwhx~kIEkYlSbbw|
z9)9^lu<p=TsNNTasmnH@QqU*Z+fj_kn*8V^6b*tF9T0oJ8J?O?!||>(xVgj`e#?e}
zg4$H{EcVB-?v0T0`WQ@;NycWq>G<7P6>tA1jlwOgW*#sMuS2Rpy`Yw8@Sh^q%CpG5
zDO}{766U!y&BF%T$Ggiv04ai(VA<9kbnNd=a)>*dt!XcUw)PYpZse!#_4$l_oEQj3
z|KPQC?qu2o+6l*19{l^dc*kpm=+Td?#<FBD6Ey!i8GqEo3w%{hPk6>tg|oZR^prBn
zz$PraU5E{v4DnOKF1-7|6<3eAW0R9E7Cp|yCjRB95wjT&HTdA6^yTPw+X!#vaqx;o
z3M$5VV&GXHj8fW);a@z^-A5ZgU0I5gsj4W-_T=uZ^YKoY8J-YQM&YUz=+19})1=qp
z&RkbqabhmZm^fp@i7~i!M*;_CFT{nnSZ*sS9?#h+;2w7=<T{>)KI{8Xz_Mr_@^R?D
zN)7Lab_2Id9>))gVzC^{1aC+1lP?GJ=q%odu?_z8_(_b5dx&(!VKCotkSHv%r^nWv
zfZ`Lw<mKsyM1j8tZhxNz;1WZ(-PVEPB4=iNat*npaFLidIx%uf)uCk3YT7ctlyNCi
zAa3$i@L|<T=+LevEpPehp-u_fX4=lpZ!;zmdp((ZMQ@o)te$d7!=5&ubVH3(mZ&v-
z2Oc_;f#K|1kB&8%yJ|fSmF3|rwvWEFYAJSFyP(r78{EQvU!SA0FrqdN_ZuF?WrvON
zjg%+ONRdD#(Go<XSfnaesIg)NW^SH?59;RPgh4$nRaC|^%MRd&Df2L6KR+HnA%-E(
zbHF)xBmNdNLK9_g^pNAhnZvC1bkPwzI2kyg$$Ga|iDC3mAABkiMwxGVm{EQo3Qd1O
z)}GxsVW<V2LDMj&QW5jJin%SFig;F01NqWfzv^jQrd_@j0)w|;^UAA`P_&t@%Lsts
zn8#qNITaV}GJuv1irDo$gV~`XOm&l1z-kj2DnIg($+#^|6b5eTnuquDMrRnqgz_P%
zjyccpHm{&wJF`HkkWx<bsSEmnWvzb<*sPy(6^#DRdgjFHQK~7hh)%vs!7_0Zlw4Md
zE$N#vG~fts-x!NZKjSeV<Op^L7oufl6Ix!@!m6Ay^e+v@Ipbbv;39_Z=XLPws0F^B
zwh_meV;JqQot;T#;#h_`-sw4rOEVTBqq_jNoK?gPlO8xFI~|WUK7xPJ2AI~B0~zWI
zkz+Cqoy3e$?V&V!)obEh)3>m;bPU2=o1l0p7yqu}$4mLD*sC%FQLF-N?f(MLb~yyQ
zhrx)^QWUoDfx<;9aNTqq)Xf(`=t=f}q9B8gz;cgI+`v5L0C4h@bh*WLFuwF7851uA
zo&AT%D~T5{pXz~_-eXeomh#GsCZJmV26s!%Ub@X@1A8;bG93orh~7;hxc+AmbqRh*
zHQ&V2?mfZKap^qc`YMYyxZNW8zlNB)S;h2zEb9-7FlX*e2GJWyDd=RBg7?L}usfWv
zb$u*4II}wRXguC+%f;Ex3h<$y8xHJPkNf}6JaHqiDbx}t3WM-_uO;#yH(<GS57w)a
zizRy7@bxWI#0MD|>!yQ?%#Wb?^(~kxEQge<hL_@I;(k{9&aIe%!sZ-&uFJuV=9*~G
zx)rFaEpFasfzeP3C*B-Kmso9#4{nA{X9X~9jV~^CU_EGm9l=s58?HsSgLQ%(gk{!3
zr-LK$JN_RTn0*F}{u5=rgT>HxFM|oMSq$oy;-IDz3kf$eK<=m%X?D2<e#h=Y(z7^b
zaC0QhfApPEO3R^{H6|c6(M}YC6uDbMO2J>Ij=Ls|)%=XQxcb@exVjH~>EvV!%~<n_
zG)y^1q*Tuk|DiH&a$y0Oo2gT4$tMi&zv+~t|6V6@dmerAMHl_fCL^EC4qWg%3779l
z!<?i@oK6v?PUc{Av@Z??ncxt*<NgOJc+xBm2cI6r%+9lzc7HLNNU_{6%S)J+PQ{{J
zU2Mq@#;#N0$cfWH^HCMN(y#|t`VbTnh{FUKc2=h;hyRTJ!XJU17^38g2HnnBWc38x
zyL|C1=L+1+w?L~Qwg&hW42+NvTE@=A%HV}~dWjxRW{9GnQY*BS1p!x;&10Ii6!p@b
z;GoDmaACc&VskHoMyU;keLM>yZAYQ^%?xny8DM0m`GC{T=gcF;JTRIO4`2Fy$@~a=
zYNc05l=iv6gyag^w5pbb!~}v$<};okM-aX*m`&%Y8PTkpo8Y*xBfRcqn8`z>jE>9#
zVy~&q1Ubmk?tAx`nr{q^{3pQZ4{6iamfMN-xoqa$iX7~n?uJ`K%uw*r25kIXfQ3n`
zap7qbEV7TlN|jV}&pwP+jz=-R*$?ZUd9eMKHBN|}#KT(?+4DVu?>!Tdufzg(@z>&}
zE*Xsd>W(hIa`46hLzbsF&g!rG@L&!H+a(>bF=7q+W=ddmO*?3va6~st6Z~8;$>snV
zVUML5+OFWo-kIvC(^1AU?KeT|Z6K_jk`6NK4ugawAAVF&#lU5AaawC3Wa@te)ksyu
z;e+7MW{Pb4mH@l|orn6u95CN`AA%ljV(QPQkdmHm=%`X-WIj|f=SH1@d!!Ub2A>hJ
znfu`F{5o<n?H1K=t|DK}x6y%FrKIdJKbH8~5kYH7BKUz2sfCZo%H%@oY50Iiu<>Sn
zRGX-$$R0-0yaGI$qQKqb6xE+;$Q_<G#+2Fn)AdI!SPpqJu5nDn`}2+Q<;GH+ddCsp
zT=&9<LLRt&o(DQ9x?)SV9;TTbMA@fi_&RzI7F64#?kO%d4mo4-bZLyf<c0QwdvQjm
z9Y%I+$C*J^EZ^^e>MLT=ZSw{+zq%LSZ4pPkSXq>`QpI(0|KQPSX$+)hICw(=*J`U_
zlFTT)AIpPpdONY}k~8MtiU#u~POM*}6IRBQgHu*L9N0bvoav&tERppxxcC$6*fFlV
z%Xi?+Y$JbkW}~&vP1yBD7|p3U`F(vqagVnGj=_<Z;iwZ}7hl5qC~ko9r+nffAPE+I
z)4+XpB$Qm>FqIbWP|{JtJIqAT$lQWfvspp3vQZXX3ODnt*7uQy<)Mu6pH}8~xICy6
zLC6pLLF!AlQEexC`ghS^>i2UQ%|AjI$Kidn+r$}zEv#_=iUe$mWap#_3FxW43zLS8
zP?I{NM}!5wHC~0@GwsmV&;$>xTE}`M)$!V)9Q>nPg-I{tk^iX!4!LCDssvMfRiJ^1
zgMhW!QFzGT3;$G1N86|^Xq6$y-Y=_><GB>g3RLmo0ZmLBBe-SnRIIOI#}aLGG2rle
ze6UX*J#|d*(5hn)c=Q&?u=DL$;ZVG+sf{DkK7**oBJAiC$Cr0a@!HCRpqiTw7D^rP
z@!ARKEwV(tYx$t4oejJva<tV`6Lk9p0oH7W$m4~;!yM>~oX%`mUC4yw<uiKh`6t|r
zrfQ9Tka6!7Bbzyk2(*i!!niaH|2#?a$!bQz+?|>xu3){w$H}^=5XiSU$TVCCXLQ_@
z0V-pdLj&rRvt$kL_@jfg`}`?tcwQSUP98y#U8z|5ClZHMBGJ=29OIX#py)eqtbcnD
zNlzq}O%!76s!ZH-G8MnMt-^PMOHjp&_1z!NLf)|?y!KrQOT}Vv|BD#ham*5h6Id>)
zXEheH^Z44zBusm>9z9r%TH1^CjeN~TZo)~Hi8_KS;+LRzkv=vk72pweCRS{;A8l9u
zgY(jYY%Mhp^9ts(K1z99dEO0M1r1PE573d-wxw(s_||TQ?Q=d7#Y=C9qQp56(~-q3
z3cJBtXBv9%nT-|2Q}FXSaVR_Vn`Fl<f_U1)sP0gqJAaGeK#4D--MAD+Ma~f2qCIpw
zJI@FIUWOCnt!tmq&0JGqJ>%M?q+P9$7Eafp0%JSDVooNQuV6Ea+<!4SCh7Fs)s<Ag
z__uDz7a@}A9Y<3>B+(EjOQy|g3%BCi24qG8v3SxB6K<`<1uLR(v^fJq1i~;w&<hvH
z97SJ!ZG5Iwg`=s-D4G?BUhO;3x^_28TZH1;P%|8MKZ<|Pn4)oM7CQClpz(9oPbI64
zF9wa#%|!v_bf)7zCzju}h{RRNU!X6g7oM54!-WbvoH6nbvQ`7$r!FY{-2rdR{|p~l
zPrmIaAMV(*6Q%y-kbg1TK#b#!oFAfy{8wQ0k3;azF&X|k+=XYK6`}gTB-C6PWAf8;
zxO{9jO9h&<3~2{&4s&_Ki)6T8hTVu^)oI3^Sx5a^Lm-B4KIEF}Lg=YHXgM4PgRS!b
zpXk$n+X_MEcPv#rvXYK26r$!)&P>j~4u)5%NEL1<LeptC>Y=%ns?_gi5}qV6Evf+w
zpIibHd0CRH@Y$Jpd5ez*hVG$nUzp*$J8}5Xp$wHw*I@C&eQ2Ax2@Ue2(epnql&@cb
zyK@5Ys#GTWiHD*(d!L%E^h6du!Le=YFnVz^TACK)<1Nv+|AQ}9*TtY=brhyyG@cJ|
zLbKDI(4rWEDI5(PY)C@?{|2Ga%nVytze37URwIbiz)&?YG+pY3n?kM7XNLpYw?2WF
zvnD|)e+wEX{eZ!pKS0&J8wxt5P>yBF)K%2M)k6eVx(Z>w*%XZEtcBm-L`dX&KiwLw
zequi)fSp2@iCAMi^W?E4azdk_P3jfV_fiAS)u}DTefP-ix0bZ;P6fj~UC%h>4l-Ru
z(a<M<iI<=_M&!>4(&%CjD(;j+OF9ViWv@T`8yc9CzNg{jrE;*|A4eL_d65lgglSt`
zMk_}!fg2HUf>upUrU&9rQ49>n1g%hvZ_B{TCo-_EJ`UBR-7vW;9DkLv{6l>*USE9@
ze-Ed!9JwDl3)bT52_H1Aa!0-JbGT*Ue7vxQW$#pdaBQm^7CWS(%PCKM=C&1eN&uVl
z@^IFpow%Ro;I(Q4v`sg|j)Q%mHGUDA?d$M!;(Zt?l0mva68CxL;E3}}sQV{`_o|P9
z@AO94SWyRA$8N*9-k-qtEE{4PU9ss$7;2xiXInN0C>(lD=KHFE%kvg)##|vRsk+B)
zR-KAFa%W<`MnB`+#%4Qp9VI$n;+d9|qde`m63~^Q337Yx!@A3cu>9mH@Vv4F2GZ>r
z!P+q9+cAAAb5<Yx6=zf9HS@vPFp_lS2!c-B3dUCbE3rAr<^+gT6A8gT3}@|krkXE_
zq#S5vJor7RtZxsKBC(S`8Vba>!y8dlI0kcvJ&<<S;?s?}Sfdh$iw~7z?u9Zed}5EP
z_Wn3yAO;m?AHaYq?rcth7B;giw>hh0FTT1BZJN|^2HTTuWxb+J1q(5r<-Y<tRnbgx
zF^Uhzpn3N<xI3(kK_XkQ$0ryC{XEdlPacO=595~}WsKaGjD4)v*PPWq#<gQX{Du+M
z4zes?{4AU`!x+`?AB5}suY&3xZQR>#3Yv>pJ?yy&WN%<VET<m6D}R6iyu~wIF(0p6
zM1$wwy+pZd0sQtxMmB3YJk%e8r}oOUEEoVD=Mawde%-xC6QH;{j(S~<q9Q2?Ozk&0
z_~JZ4IzE(8&j+!Dy3B+(e~;752XWw`vYf`&rZLw4?lPTylo?D|O$DzC(r-1g@aTyl
zaJTx>1othBL=r=`#YEzl%qT3l8Hb$;*{nX7j?o%v_^&4e*WUEP@EdFKNQe|JT$#!G
zMIumsUnJ{;G{wVp{^%6F6x~yIVNOB}HuIU{$nzBJs?fr{zumDTYc2}0nFa;JE?BY2
z997IVqUY~wXj(0e2@hD0$omcGxVRo=Ww>}RZ#lXS%Hg4?>oCh`Ar^ur+7@$A&B_iu
zyw_t`<rrN4sE?gv!O#}^m^&n80FH-xiRmdme7!9hvR|x+F{%R(ULFV8Kff4{24U1Q
z>?Lk&U+OZ=62{G@@P-R7>FS+&0@v!c1G7SfezJ9>9a+`nUa1%z>M#J8l#7f@jv{1a
z#ZZwSpP0bd>o9t=fC!#VAR*@h7zIURCS<`B=xF`L2<ELPN}J8-u-Sax;y4o;WRSpQ
z?6}LV70Mx<C*q0F`Zz3E=8u2p#Ng_L!?<qB0rc2uhEEfV@cWeGcy@jQwtmRM;dN;^
z#Uul-O-n)kNLMuPvBxYMOEh^Ag7Ry9@p8#Y3}rdrK%saPR$YU~a#>D7FCLfmhu}M{
zc^KZf0j-}IV14)`G?lSgGD`n|x7QnGXKcV$UzSa2WBH<{R0tdz1?{Du;it|_G@r8#
z+k*;0e(VBVdRqu?_x1ztSquoavo+%H4D$V}0)#!wfaUvC;Lz(fc>b;q{8sWqq*^9)
z%EUp+J1w@)6eb#O*LjUM83-$?CwHqVp?!KRG$)^;DXkr3YQjty|8RmUygZM-FpGrA
z0}}M)iJQ#tu^2M&D4S?-dzsD)yGe_}49F;tWb7+?n8{UD<ags!rmc5BZ+O->@>4pU
z{*2vA)#@)ZN;R5L@MIhAn#{w+wt46(X^$}mX*i<13CpFd@ips-%h<Xfdw)2hL(3Kv
zx|W8E!na`XjUc=l>VRUZ%TVT~59YbXV8Z(~X#d^-%f(oo-z^EFS${(l2BS>7Cd&M_
z#plrx?Ayiq$)CmJ=;f6teQ*i}3?^gZTXigVIf5^x_u{iF(x|*D8dpsfMv*Zibn7|>
zhNsVfejpFj|632a@flFLE(QWw#)L298Hg<R0QC?-(7t30&rIh+`}#K|zA2As?x-c*
z&U;8tMiH1;r@(kKml;YNVj`Db)(uh?0hgmYxvsS_G?MR9tBBw^rY};Hnm?3e^rwsQ
zRML0RiPOtqR6G(cI>kd!tuB2sT2ECsD?&qwI&{ti5_>=fu1y-!$YuH5>I_GsC$fs_
zTkA5lL%wt@qnMlz^2O^smgoLvjMu6>ah@ex(=SWIE&(T$otJ<oo0V{La09LnOF&h>
zFm}$i4uggc;K~<XSnHmO3F0Z}eKQrMZnJY|7YAI$o`1xt%~&_p5vOQ4<6lc5EWhr9
zv)j_J;t*Q{Kas{9?<`DUwUfWHv+@0aJjzW`#9sadD9|2;hhQq&)jomnk!PUh@)W!-
zFT|4NC!yoea$MCN1NxF}a41|Dy^=LCY8jhHtAC&TW%Eab)da9jV+_p1I>3I@1!7hH
zoJee$M&vi`2H}FQF!<{^!+GGNE8E`9)a4a{G3g;=K?yK^afE3%nhpK9nYP5thv7Sy
zcsiL{#AOv9-J5<DtQ#`8{#7nagOD`MS^b3J*g3cQr_ZMup9Sgo$&XB5ay0d@vw)CE
z2|7GofoahEL>?*c#w-zSeBzXfzdst|A-({7!Jmk^{kdqvdU8}>9zienNIWd+fx%w_
zvCuae<qx={<A5h#lyJbs^Zc;sK^$r|q~d~VZ7jcX6y>9Qu;H&O=51P!Y9h<gfnN`$
zYnNj9z4=(rYMN`7JpjIPYh1=Jg{gO!;BfjuG!15XRkrq>KAj(H`nF?&upgS9ISxxa
zwqoHdIc%O%104ghDD&_x1hScy*S7XR<aZemyg*@bcRM6b*$SM?9^}g2X(+XK1M$2p
zNK{<5GVV!hL3U~h+2+{>?k%F||3ej89HpS-@mBDGL82BM2`x(<LEqAi8zNc*V(A$W
zDC7r^L>y@QQ)_x7_7!Yp5^2VNeY{BBWS*mnAnnmgrM8{9M4>s58Ey{Yo!s$|X&64n
zbvqMJqfbWCw9~#wCLK}mSq(~>vOM2OJ?s}?ePn(q81p(AlXb&!@%r7YcQg*==4Rmi
zfh;^@vI#poSO$fiqZF|{zf(8s>o~IruSvS0@*^EwKYJBUpOcOCN;`3)W)6lbo8VgI
z+2|1vjoR$YFl?t7YH*%FMWr;xZ;wPlG1kXUR^oWMGR_wt1kRV07_He0(i<(|vd=vH
zQZ@iBm0#E$0^(?VUmcyE+yVb--q<mZ%?2~<h51?%c>SIwPBFU(d&Jmm(Ie{kNYWSl
z<kV<->~Y|DylC~=R}O<>N11Q4SAt8j2Qd|z3r-`=MAK|0EPg6U`^I-ue}!Gl-|f=S
zketdJHn3uHpV|SZG?PhKxPrN}bpiAhoacU7FHaj1OqrD-QB<L13(Ycgr47%Xal1DS
zFtq3)<FCAlu?u1K^-tb-$T9|(1@2>gB!!sz@F=?6OUH820^G&tiXy&V_%t#Bi;Gx~
z-h%|pVrTZe%Z7O1S|o}c%)k|=SZ;O2VSK;U1`AJb#^_46#%1PV$VMS7D0D*pOF_6e
z-wWT{N}#-|FizRR=2BXU;fK5xn0I~-w!PniihTdU$Wv9^@uLs!SO{b6964+}-2tYW
zW#DSl2_;Jhz%S4fY*jT;m-WStszwv7#zpwMU<8^&tDxT~1C(eX&=(7Uap{M*dk=$T
zL@(U?9L+R-&w}FG%fxS^F7U1GVDH*oCZ*ytGi){;^tpd@$LC}+!z(*@)oP+}Tx~wp
z|Bu7$@U^F%u9oETy(6Fy+d<Uk3^FMbF%VW8OE}kUd7NqWjK0Gpss1cWW2<W!B|c4R
zd$fglGM-J>y-h%s_6SV0^G3nkIJDfHgh3yoFn(1iZkJX@UBPnJGg*!MEcfF_iD<O>
zeFPmp9Kib9F?h);4=dGUaJGRf_FE~SjFbj?)tF<*yy>XtuYpB&thRr}8(+80!v`l<
zqrAX9@Q4n<`dmT$Y5f>7zdPd0^~=$uaS85_5XR20%kkMnGPZ|Zf$GQIV5w(?jaKup
zB{~U&GhV{j)H@(2podTDEn#y^8mNE14k;H(h=z9ns5HcZSE(_Z+2{;bN%62vwUX^S
zBjI)DdeD9=%W|tbz`n$n_<afn&&)mKr{Q*5;=Y|5b}ou(W7%9;{{X_7wwW8!Ys|p!
zSTZ~&N+6($*x7o}589fL@m7>kn|+q)^v)zhTBfw*PZqb+WF<9Ttqi=GCa_^FmWurp
zq9e;p>40tky0iMn6C)?In_#v2q)ZgA@I|lDwJ6ZT`glg2a6!8%imb^&#vm1a4GQq(
z-*Ajt8-;y*{`e?g9YKCMHfZj^5Eox8*4l|NXH4*}g*-axDPi?(J(h{yiJ~4q;U!y_
zmgK5nb=Eu_Sn?gzhqvIaTO(i{dK{D*M&RL*b!fU$hRyc!M2JpCrQa!_^G*>zu-f*-
z@*()JU>Vk{F2yg;pMlF?A-w1~6E%X_{Kk6>tS>Buelr1Zy8e-H%2RaB7ezpuq5`9m
zw}!}-o4`poL2C8&7c<nZ3f+Dyp-1c(=-=MMli;l+d=Yh|Z-Fo!5>lX#cIp6>R0HVT
zCnvwmfIIP0P<vL0u1w6Q?YTQ>N!d)|e=310g!^#2Lmo2<zh3jqjba!VX)g-n5wxHr
zmx{Z@;B}cWy!vVvo`~9sa;<D#(;ba&5r|Hq*=WhW#kQ{cSnuqH$N%_Yn0W}UuZ?2)
zatYjSwF6-}7oo`xUyrRrhtg%ZX{JB&vwD)Oum&Dq?u^mP*W#YpvN$?V1J5M<hDcVM
z?F{wCPf}Vq?fxEADd6Cv4IGp`J_Xa8EU@I)FeDz=#FuY2WB#GlXjA$CW*(>lNxjn`
zFQku-hCUGQwgMlX_y9WY!{8I92`v{2p{8dMm^H<L{-0)D*Nxwx{ZleTp6DPhHK|0O
z-<O!$bAe;?x268iEIRP&7c`78%$*4<C^K9D8B^ULFy5L9W(!fdXiFHelY;R<ZD_h7
z3SUmg60OArv|-I9CZi~o>%sCXK3CeA{SqaRKPW@>+GA<c?3uKB{5SJaJDPR|USm!L
z2I9HfE+{Lz4f&cPFwkKy*4&N45SCX;_gIg2Ui;!t6(_u7t&N-7He$EA4f<4YagyVY
z??Mw%UT{A)PxZzf30v_cn?sSCEP*pC*%_3M3T{kWgpzJzC|R%t2VZOAJE^UBiseK%
zUE@Q4M-dFYs)jPVufxvrCMfwJhng$4W0#Euis#qBfz?**cYh4lelWy-9ZggaybgN&
zwczS@9K7t=jETHt$eHsQJgqCp*upc!T{0AuKeR%IC_iRZ%>hC7E{NqG1eN2SBr>zM
z^~;NAq|Gp`wYa2_WT+LwSZ5~G+p?VVgM*+i@`JpVbY(b&#;q=g(zwG8ZM+~6HZwI#
z7w#{+4PnkhjOo`{D%dZ|gjkp{bORq%KeiBRfAtdc!y5GD^hqWoc@;4y3uwym9_Ffx
z6OLpWV`~gs+q2)_2MK))Gun=BXN@q_{4m-jZbhT@w&;GBWj;AxxJWq?XU%nCbBY}B
zK%pk`EQ3)xayvVlR$#gMxhUZ-hNH&_76kg?DC_ChleA=cw5=FnGzXvetKs&48u+JA
z4^3|?vznCytMQBA__<#ouuqR=+u7RTmkcg<Er-C%{$L_L1YK_<A&%b^j|G2)k8M@Z
z&H8q4OFssy7zS)EjKFJuTgVcdfeKRwxcbWbp~0h=%UPD*>Tj?gIDWz0v<>3W_xc^H
z*&(Pch$eRV0LI&sNXed7h>h3_8vQT0wR4rB>D_4%JjEq02hI_Nv<qarG7my_1ND*r
z%M4o-b06uffF0)o=_7kSY8}c?A|rOSc8}IF{_UBJzP2<G8<D2^(U!b6QD@%G`f%3k
zb^vvq_o9xj3)Thsv)Lh=S+DIpJoIHXhRu}6i(8gpyTMu<dGC%R0Ty_*Is$pc`X~?*
zgZ}#+&@y}-270qS=_Na?W#`c@N?|yWV}awhQqkJm8nd>q$BsyLM$)?jH>hi&EJp~H
z_6wlO-Bmc%+#S-!*J0&nS!_rXL+v613{TC2i7R{XncW$vmaTwSN?#y()<syiW(53B
z4+4EFjDIwFAS}Kgt`(_3V2T!-2d)f*>oTBi_eM64I~~L=UVw{70=Y+LgZ<1h$eo%E
zDqd@0E|mb616P@x<t@b7Zvqy}1Ve27Us7#qO7+zQ$gV?$kkgq+_<o&dJe&_hd?EuJ
z*=szl3u#2^Y7|7aU*T$}a;U=1US`;VpVnGs5c6sd6CWrK`j^%dYr{q4?`^hkC)UWZ
z-H#$HuVBL2fZ-{A`115xJfE9^O)acHf1?4Ou4gs)UMrM5v=Zf56O`l<Z-j$~&}Wu0
z2C%czobk=9UVj9i-jTykUTbk;8NqT9#N-@lv<ynY+_IJUqzEvjWH#$fcR~3Z|3LN3
zR~Yb>!1YpV*vvE;Hs9bB=q7i-i|?;N_x^8qTCIbxCO<*e(iHeAo&pzZZ^2sVg8bm$
zV7J*FWOb&(M$!VEn?{J^?Fx7iEXw9cG(l1Wz&a@@lr3t6l$sjimnRJpi!7Pp*ncGL
zz5;w)H32H~wV8uk@*%QFRaaw@O-()igb6K;hlbULkUw`d$vF6kE6kIkPG_$(=5y2-
z!T*Gq@sOKLwf!1K!{aemPF4(p;5unrww#U|U*u`tTTVHrrFBV54*T}6Lxs>d%vOoV
zVOb+=U$PgCQXG)?!5U-JbZ}^e0fxQw#}D5ZBDoieuOBQ%;gb$n^}rnm9>-v4c_I2F
ztYMQpSRd1R16(A^dVdq`S+Da7To}PYoNt7mY^+($iVvGLHL+!jBW7mL!#B5<<I>M|
z7?L~!B}V{F*-VQr2?>0W^a~V<#Btq{age+%jLA>u;{n?q7?{pxdYn5Awx%UeJcA3>
zJD<RP{&R4gZyUVYpUCb}xC~$Z%)?uk8E_gC#qNh9peD&ced{t%m~aEz_AH{cE1q1C
zy9blrSBd1yEO6hg4+*Cag0+4j;i#MG4!S%bYOmO5|Ggu;ygA5vt(>Xepcyo@GK~I>
zR<20*1rU@AX4F%Spy3sdD{?iR(Ob-NXKE>okNHC;cEplO+3%pCS!>bUI0nbrI*4cC
zhJMkeD0e#xRW2PyAE^xNkM}{&J2(8Rz8d#?gtIl4Hg2xV#+AWr=D|m0jQHV%@p9Yo
z?&rm58Z-~xxT|pF+$yZqS&WCpwbAy`U+~Z~$AKYZ+-J}Bsfz1SV$LR97|qrO_3Ah$
zaUKc@j>9N-I_q`g!{1J`v0)(xyMEt=YyPjGaqnMfzom-O6~&;ng3a%X`UV1ZD&W|c
z0>>g!Aku$2xGivp{e0EnRjB}K$(up#hZE2fj*y*~2ufd{!Gv7`Jo+OJ-dkItZcY{a
z*;@p$xs}YvOO-5d_7|4TjE5H2Y@%~Ko1{FgArfnKnSA~%QnFTpCRDQXzQGZ$$n$)t
zRurZ=JBNs$z792Sp2eihjV4x|&xxm(G_@uTJZqJI3|Ub~(_O-_|Hm4X2;Yx${Y+4t
zWfq&A?Qmd&2#O6a#|l?-ygSDXzeXqFyDO2XWoCe@pDUmdTNCvi(!#f?i}8}06LwEk
z#NWRJ@Vm(p9N51aweDzQ&-f~=?pMdx)&iJ3?1%e4=-`GQ(zvUFpxj<DlpC6jcUi9R
z%~wq<40#7)M(UWk<P#XpYzEB*<KUW^2)q9413TRZ;CICXm6ly0(;8Ai)nzJ1v${dS
zj+gL$bsIc-&1%|(-@(UC1=>VpiJ;&iqQLGJsuXsC;k|F+6lBBfpOL_cc}dKF&mn`A
z1<d4GCr}yig)zB@<j>c5IJfjVY!NO5iL8FE!UJt`tfCYI&sH!EI}6EZLou;-lP81x
zu|&hjnd$ha0nNLrN!yG<(o#H~IIWmRQ<B|iq)UFQ@ex1zaD^_hwa_JESJ#r!IrhYF
zhX?tXmrR7e6%uucbfW(DG?}ccBZfP|$;0pr(toms>|T716qarzcQ39Xi>~C5X+b^Q
z@Fr0bvWG)DpB-xzJYC7{%xG@q<nL(-(Tn5NT))B-6raVFY@WyXdsXx9ZU4@*eLR<O
zvC-qXEV$e{K2M9ev&4t-NXuY~pY=0;T%Ix#tUpXcnnQoSnMY?|Go-&?FQi|tB-0(8
zN2vJ)d#XCf<`R|U(XN7{^z~34eZy;|U02W0OHUf9w#_yAl6RTfYjslQ{5btNbAWEz
zI8Gy$@?CuWV3O_=8lySuU(yH1M(ELvpJ>j*XH>lZ1O0X6F%=2BO+O{|(A$1JG&1%&
zb({K__DgqCNtZ6V?$ZEmcs@kER=%RWLyu`~#B-YR_%&4^<MhVsae8Ox1f4iPNxug~
z6YIDv@<>0E!1io*E|y1pm8(cT>*KD4G!o5!mi%-rCd;zRh)8`dsoZ^<oM>Ul$H@Xx
zQgMZR)Q=&?iHdALe3r~9+D<C%OG%BgCE=M^5pA(TZeaXIBJuhFce{HzZ@9UXTh_rL
z`8jG_{}qq9ZJNJZHCnq{8<tkI4m*anR&dIgly4Up)c{H6PP06{@Z}UU(x*f(7;d07
zKJS?6bpo{BZ5MUzh@!5_mUPC6BeXlYh<1vn(EW=u*o@bFDmGk1kE++vovml+>CtZb
z-<#{SdHx;RcK#U+xOAUF=~H^<`3Rlo{g_I;d`YJ+?x#9752=624H^+LOb7p7q4(|$
z(izh(QU1AIw8j1zRblsS46pr<`lWVJckvP0wCMqj_j*JRDNj(-j$x{$_LjyPeW923
zexuR{CTPh0vqZ78k!UBE6Q!wF2%q9h^5nt`;wX2MR6qJd%*)@Bu_%7f|9pcqwcjS(
zi9vGe4x42)zmZ7719Hi#i(LMAgnTd@C+_TC#2x!KlPLi}G*9j$ubqyP!mB03Bb<;f
z(n!7)A0|O%f4PoKAhFyPLp(+H5%&QO87}|KwVj#H%kQw{u^TeE{+nlVBhL)!UizcR
z6se~%)l4W;vn-2|9?_(=elM7PVv%%o-!|G!*xcuQC+fp~yHa-&XzR8VT4{fjwyemb
zTZc;NrNAb-u(FZL4z<zMMt7-#_cfaR@G3PQzfHSJp3$wk9dxGiBWfl4oXU#bp~asc
z(@(=4^pt2n9kU#!Tg97cw9P%5;&G2&`p`|&K6S7jRMxX^@SKjm=%aM{L)udCkPi9{
z&^L9j>1gZ+dS=518uM+8%(weT7SUH^TAcvvk9$d~zl;;!;wf<IrYH;^9VSUyf{+_7
z4H_C>h{t(8h_29psl|h2bp0=4KYy4^^iPASQSV6<IYR0N*&O1}{}G2<8_Dl*A&?Ig
zh8L4Svc&m7KYWl(?EONro|+KbbqZv0%vth$T{`hR&E=U|o+DmHM@iRfbEcNBl>4%a
zFlsg@7}Ix2jDFqg){Kd#%$g*YEotgxR089un2;77-=$9Fo+#4wQ|xI?mnmJk{|K!Q
z-$O6XkEJ`F=2AhGG8$rEM02OrQ$v#~x{{92;IeMIZQpsSwD3IDeDs2bxV6$FN6*t6
zIajHW|9_Ns`aD(k?4{55R#R?v4HX`$r%_pLG|94o`Y5;4sQ+4M&IG%E(ygCXx7?)%
zMBmbV$<Jv)-ver^HB6ItzN9<;@k5ESFwDLo41&&r;6F>A&FfcyHKmepH)IJ!7HYud
zJIkQ-+Y(56`;qu2N<mlZbQu4#0k}`40KbnDVdg7o&z%SV+37>C+bHSSqYgR$X2Fl~
zE@E$Umt42KNiI%K1xNR#pt(s7@>j1Tzuzqab?+f^Bj^D!|6<#Et1*=vdi9dzZk8mn
zyL`xab(e0Rs5dk4XC2`L=(MzcKhCHu(C6Jz9cMg0ucAA*1=CAq_nF8EciQa4rK)Kb
zG{jbmzO6~1zQQGRoeIl6{bRGZ1{<i92D?M9J(C{mZeiyjL$t=Of{MCVQM>U0DzmJe
z?k_(>4VF^sXLXB?2-HxaKM!g2uNs<oy^eZWl~4hdlXO+(8CqpiPc2V((Z&oOeN@m&
z)o)W;w&@f7z5ER|KKO#>JiALB>RwV;W+AAZSOk*erm!kW4n!NIK&W09q*`R)e8oCY
zn`Htb5wcLjzN4z6n_;TE5riCYh2CB@Sp0P}(A^tAe(DNfeuzU`uq8a)r2&7&tzq77
z7r1xhB#}9;3p)L#pmB3L%n-K%)0G;K@nsrJ1Oa@_<B_I29Pk%-$X)vpN!il`u4vIe
z^1x9HJZ_l~;mIx{*>jBqC6o~JxqM7pwJdko?OSVtyD+ViY-PSW{9v9gzsVS!s%7;0
zpE6e-TGKB%PV~2N2p#{hnO^*BN&mJSq$fT_Qt92{bge@))sHzrGZF{rpW{9B`>i_K
z)KN%}Nw9nN4EosrcpH6baf>RcRZz#rm+1zfqf|DxgO=spq|fFT)1TxF4gcLwhtK>+
zt5;m7zZHAwm|iFCZ@)r4MX%Bfji)qh!z0?Py%I3o0oIkS0yQB^xWw)mn;26BEr<0m
zknarXMay7Pas@nZ+6T>l!eB{@Gc5Cp1cOzQu-nlYBqD8rjgf&LpV_`|`Vz2+6@Y!0
zok5D_Z=`n}fW^aZFtC&D*?p{`*IWYfe)z!p{%RstxenwC4uIj;65=1o2QRO0fID{+
zh>D6Jbl@Ap%QhiNTl2`@nE7Px7dIkUc#Vm$K1V!HvFvPx4DVq?9TS^4kCFBCU^I$F
zn6S3(%#@U040qWFM%9>~^0T{%Y;MWYTOV8LNWuZ?cYhsiRZgZ$`_t%!@(kL2q=D9M
ze@OAiMfy9Tgqpu8qA_rc-jldZMW)@RW6Ms_{Nw5Lpm!<t7AT_PvS+9z_al{>bAj%B
zdXD;+U!(^QcF_PKN<+ra(_WbYs@KfkyAz!>xpFPk|I&q9M|MJs!6vx)(G?<VH?n!l
z5wI}87gD;GftOV(M6HPjof-L{^J@-xxtM|eSpaEaW61lN3en%vpjXcoG_})U13Ap@
z_AG>5U9QmoEDIRVSg^~E0R45F;7mgx=zL*)bmcoCg}WW_+kVh@Plx3f{UG*?85Ftg
zg^WQ<c#gVc&Lc;n7+uDbv@imbY3x3^#6DtmZj#%yVt~+Z0t_c=29Z;8VAPJevCOVM
z?X1@1B^^h4B(jR#wfl>CP``~ju9!>9R9)!B<{eb#;945pY(X!@1=8fJrSwBnJzZ3j
zNGl@B=nMCQl(7EtqYjO<=S~V$%^0R)=6C2Fk;^oH=6`f#ekS$YaEw~6siV7xE9vX)
zx9DM=YgFaXMf%t175#jyf!61DQ^}uw)OvO=73SXywF$N`)Q|$}`1e51ja1N$j)C$y
zsc=yw5QYo(fsL*gOg|I~yUvF}+1Yex`IG{B+)U8iX#jWbMZ?yK0(hzK4PITb;KuI?
z|4O3ZD|aKjAN7J&d}$CKn+m}Rt{^tw46F<zA)q}S!lG&+Xm&b$7T*aU#{=L?kSkk%
z+Y;A+e_Z~?gCOvEA7q*-!||}^<f4iwoM(5))i0P2KF4fHcJD062szEfJ{lmk_C4Il
zrJ`JR2Ln%gqJwGnn#;^l)TK{T7g64`4902hWoql1MYHGV(bmr#`oYVIS_-aa>$3nl
z7^g>76y8za?Q5xZg#+!GOrr*In<y`^naXtUp`E|;>0Iq7`q<<k9TP00_Ah(stK33*
z`p0qVq<Vt3tr(_GvnXwQGD>rkTj`>gZ|HE&MSAs&4VY_YLE?A>*xbp5#D6hx<U|zw
zc%BFzJeEa1nE`jt?_)LeGU$qGfvsvmFzr|qSVo+L{%c3z%28kVyEh3QBxS+R_l;28
zeUjbl$TBnhHQ?KS45nXlgXkx*u;YR}d^}waH?5N(UQZP6ZC?YW<LnMZ1qYB_p9U_K
zGuSgTgUUnRP;kB+9%-Bao8^JTCVw3~r9NO`YzPBuY(b<YfhTy+k+}QMgu@#W$?(#k
z)=P0O82^SI-Ls9;sDMH#!`W}FleHs_xpDF;<5^)qKXrU(-Ud8ouJmlD<xT78MAHvu
zyI2Y}czTy!+$2d4e9xrieIE4hmz{Jo+dmB5-#~|dZlt69yJ%*A9NpsFLF+CSQbWBA
z${EY1PTSI{@tth?{#7e=vAjpA)O|X8?pwMz<vh*W(oZ|nGhthYBMfuw;N#*V&^})Z
zt|_^|IhF^OJ!O#ZS^_i6&cLSbJb3%_2z(MN0q1@HfxG`zINy~I31c^5jq*(xo5T7)
z*g5WomLe!%&O-iPGiY~z3hzS8AxCi;B;C3O5y|y1_fZQlazmh4+76;#*Fa*=0PKr8
z4Gw`Fz&jfOu`0S?%kD#F7V5%-#fxC*cPbfQpazPQI;5c-VD9-0#!-11^XY0M6TaIN
zicI^-0=-1~^sqc+g?TaSExXAM6?@V)62Nut*I`zVCD5t&O=!7wDI<SOg(^n*GWvI}
zQcH(5G<8ljojW_2uKc`<4mc*#+_l%~(#^_L&ZvnN>#nC4Qn%4|aRa(Q;4D?LiKVd>
zJ#^EQ2x=u!Ll-4w(q|#n)S%%%`eo?_dQJZ{&CecZne=Y@cH>^?8jA&m^+}Na_5_5j
z<$*|b3cUJN%kI93h1sn4WAl_VuyxHP_#9gY7QcGnrBol>oBITM9~i^FoYU~?Ybz`i
zXaUi$^&pm12p^m4K#6-Cn*ZiQYTtSGdwd4Je>{h~g;!WE`2cvmI|AqRE<^u+=OLo!
z9?)~Ouxl&;Y$QhE-K}=exX5NB9b5_4z=g$PJP`d>3$kZ7L6*r0n5bv>e!gD|{>Oa5
zaaR;L);}b9=O2@Q?==`t5dpwAKiNGyZCr^-SKX{*hV-!dWyUZuk_MmeW$ug>G1W8o
zG8P75)Yl@Do((UhY6>A#WSC3Od0l4CzpbJ+)jO#|WhxE%<3z_s_Rs@j<#g_}YTB85
zl4_0Ipp#Zx=_VHs8l?DwYQ`U@o?bU-UEgJD__~Pht{tZ>g&kDIu@1_)wGg*38J4ZB
zgKtU~V5d6+i#;3R(Ze#h^6eqW)b_&KYi+Rl!WHO<X7`^NUIn|6C-9=78g^8V!N$Yq
zVDq&>IC=Cm++=t4srO%i#=-`0d*2In8`<}3_D`_kKMWJ@HBfN53EccS*mL_TjK)2K
zuG){VVNEssW%DL#n601|<^qdtzJrz7IWQHv4<Y6^q5IAih!0JIFRF#$rRxhBcO5`~
zVKuxC5P`xqQ{ds5bST_m50mGY@B(M(Lh|x1QoGWZabk0zHyQpToX1af51kgKvNeIc
znFmfWK7vX--#g9pMipDrn$4p{AJb^>%$JPk+a2^yyCUt_ugvD~ETH*WqSV)<oJviw
z>7rRBRIsO(*3<{kH*FVap;tNEpQTZ+>MQitk``*yS5MbfU8FgBmGqVJDIn~OSCiWa
zVr>O*?rbjj`<#PeomWs4(F9XVS?^coU086V5wgBtgu=5IA#1@02-cp8WvXrPMQ{ps
zh}?oQ*54_x=sNVhx&d4M4Z@ZKAJ{we7pyz?2Xr=$!imL_n5uscRPtux;L)2f=eh*O
zRn9=K$SQcB(htqr-=XaKF%Ww*2(lWVAt2==r2LPg^KiuK{ol9{WhE^kNs_Fztn<Dk
zql6?RNhL`tRHCGky_19_iISE|CF0y|E$z~xXdt9MDed*UzrTNg$K!d<xzGE3U9VSc
zD5cVE+H@jEaA?V=FsJi$;y@!Q7~iMRs%Sc#;Yz*_X7TsZiVm+x5<2T&SiV$>B9%R!
zl=Ni|bMD_;wEK43a+T5~mVZ>9`{-{9OIpm?!scCKsogvA+ff^_>TEBTq&ZHs=m=vG
zEi>4}{UiCg-JhA>S;MlY-C$#NO4+BTo2+z(A3JB8!Y*VqvCe+{`nkWH<r$PR%E)6C
zVaM3kRa@BUz$?s7VJB1QUrmwcs^~`aE=n=3AxbKuUhP*%XHFf-MZKlMsn_YB^#{7f
zvjfuOUQ+m^M>KzkNXE@Qbfni~I`plW+^ZB(GP9Gen?0k8o9d`$ok*ve<+wlL1hr0=
z!h{XYblRj3Dmja|@7z!Hqo<NaRCJMwmL!_9gUB|PyA6-rrbyK%WU{6&7V(*@(lcew
zm_9)1l1iMj$vMXxdg1xXQ#6{NB~q2T>86Aah4y(syFWjsk{Q($F{`y&NhOd%n-9~j
zas5fpa4X*@j~D7^Eu>8SPeRCrrIrb4f5a~z<AtW}y=o-WJlMOUWZ`RoR*jW~CsWe9
zE>^V7WOpK?#TNT4*2x##V{R;AHut8p>o?c1Dj2bdlQAqTtAu^peTzBlY-V4dWU;|x
zxSLb{Fxz-$5Bm_E$V$eSvE`BFwA%j-$#ykR+`TsXY*#_gxhwPDa4Gy-Ur7aW5^yWM
zKsVNZpp1(TD31FJo2JM?FZ(i;|Lq}p@hR0Cou&cZ+&{7FGlk22r5P_zQ`FbNC>(d6
zmLGja?Tt<3?>-o2OWu&fxux{Yp7XMuBynZ56!KDZ5k3AT^=*|yx@!X|2jo%7@<G^;
z-xrsMb5_aj&s1upikEyJ+xuu=Tv2>N+=fBwQPQ|#Qcv9m^GK><gZSQdC%y6%>GQkI
z!v3vEqN+|jDJ>f>80U|t8RIY4RMwsqY9+jw{fB8R>RXqvcj#zw{Mnu2rouxk+RMfK
z-kTna#NMi6!@NyQW9TOF&Fmtv;f4d-m0T@4+OJ~9OBS(q*}m*Mcgq<^Z()b&CcC_<
zf*l&=$4Vyixz*ngrhOof&GJrT8<Xov%(_8i)b`Qt5tVe%>?>Wk{*nHZe9JR*oM+io
zLVHr%X<$b;RUNG4tOqqb7{eJ)L5dIy9+GwYK+Ndug)&r;;H!(QD+-*ICXeZy^{`fc
z6z1v;1i1`@%=^($&>4!+7Y1OxxdG0N)57B!L*Z<Bp2`QyKxyGbSno)pf(^2eQ!3+~
zdUXW79l~=e8|iz5Cej@O=!?>Dd{5>K1P4VJ#^_VTgyS?JhVPJ?Uy$+5LnPn!grXGY
zlGe;D?mD_bgSl4MxZaKqR1}C3qkGeoYm?ZVMnCd7+LyM!sTK6Q--&M<ABp$uOj&=E
zV&Q^OIJ+=0N7PsqC>r*cVu1sn)m+q(Vy;Wpv22$nc4>$+>)zUrnP{D4pI+24yA@B_
zmVlE?E$AZqz`b3wwwz!(4?0*)`)}H&Ur#B+zf->RM+&;}oZfADO&Y(Nsd&MCQfnE6
z@k_d?adH<Wzfy$hhE96B{xQ9LTR^vF>f(o7A#HEd#$?MT+OMvF&72L@)W$nyC0g(j
zRgkqnABWEkh8E}X{H?l9VIPgKdc%0MsE&e8>o|DxUG#z!O>}#zptW#6O;%Qbp+ON1
zy~|yx+^N>Y^FAAMbuqKV5R;~iK#==>(oWZf*6BOc$p0T}1?~e5JwfC12hwnbpOknf
zh8|3-C52JO^u~M!J83(Qe2<P}PK_<1+FL^wJ1Ln~U&s&+j?=5I@4MDAeYGArDv2yM
zc$YY6^huVSHd4HIFSUBP)>5&lJxBcbREL=>3S$1BW9&?^z<zM&Zc^`D7P$KWOPmqK
zY^D-x+c}NBZrQ`y)Dqa$M~zHF?;oX|RzPO;cM5Tqfr(K!tu_Be<75=kw|OWQskPEw
zsUg^MRSs%1J1F`c&mxW(355%PDP(awz4UrWtiFTBO83W(U2W9+QU~Sr3`Xm&fsh%h
z$9H(+;izN`Sv@;+@}7;{O*JG{2$-%cfu8%CI9JA*Xg<9Vpxqz;P2ik~&B~~+?u8?d
z6|tD-S^~J|ta!^HoVOf_OWpeT)N33@)bAp#>EqDfxfkAkQ$X$Vzx16m?*4@>6}Haw
zq3B#$IJ}Le$nJNv^-VL)ljmvn&Q9@QiKlSjUOGQp4q%V$l9}(0{-W}t4>hO0s?))u
zI?Ism0pbq7av}3#S&gO9Kq0@okrjPEDpn{;ve{Mc?3ij5TlxGN3-QipnL(v&`LING
zNpT4i7X&i(13ApZHj~YWN@vSwe4%WUV!j(Mqd_Ybpl>q}_71(VtXu}klQdDM$DP{M
zjWoJV8O~kT$tp}4iQBoeRkt_ldJOSoo&oj^8HQ<IA8EueB@78LhD4zXJ_OpJ&!fID
z=WOXFM>TYG4}#0Y_Y|G{gS<aY!l7>i@X37!q^Ed6;nXWCE|bT`!Sg7L-_r(H%OO8`
zIO&@@W4QTn)b4ymgU6WS&YJ&(l$JwOu%?|frv%Z@MYqZ4`aU{qFdBy+X(6;9H_eC6
zrcr}j==bJV{O|f8Xue%V#d#-4_4z?M@>-9McEqwK-#3Em^h~<-@2pV%HNX1pRUdj<
z9wPKqmaqZ?6PBnrL5LW?Pqe$cO)NEtW{2;jva@C#Y*dO2yFWRWUCNMUCZD#loz(}}
zoDVDipVNO`{sgP9yU*es&$3(Z24kslUyO{oKxel-BCD&z;UlYqyE6tNW6*mV^REwX
z2AW{N9nOJRGa3gr_rZ@Bqmce?0wjF&a6#-3&lOW~FI^r!hqMrXb_Q~6#$dy<QD_Wy
z;>;e-LoXNszceKrmAFHHXHLd#Wi=G)jfQq>2R+~GhxH{Raq(FU=|?1z@&WGrS}B9G
zq6>B`mBsiYCMfB)L;qA+Os{Z9#{w-_e(Z($LEot2{9C$rZl+K*P@X<K8VUU|y>WHJ
zD7y7Ek`n5V3g=tO1l6t$G+e@#wt5)}tt!1ppY;;tubI%1_95iymoDmr#tU|nWts8a
zGBNh2n=tmoDbcH@m|b6<ZW*1hz2@iShisgq7dvzDrTBUw=Tf@+iG_R{(cBfpJ|$dX
z&by1*9Jd3^>cvTBBOlHN=vA-}%^yjLGXyUDn1IUN-{{_j!I;Cp^UvP(!;(q`9OIm4
z8NMHVYSR}R7WT#D%t0vTZeq1p0zA@8&}V=Eu}&KLyr;jSMjg2t^N_U86YH`BOj9?<
z^l{!;r{au(M`xoddKB7|X24_+XV_l+$!Aropk3Jy={(P4%^eor^XH(`?I35u2cl<D
z0o^oqL*CdE6nC?Nc4YJ2_xoMkneI==|9qh#m$%SCE$-zBEum+vDu@c3Agrq3-0$T>
zV8GdkFz<~gX*nbl`HuD#$I*<+JQBnBGo6{T;A4|b-RMVE5t(FOK9VI*NG0(M*z~|B
zLP*bU&e;AW`VaUlXa!GWbvKK|qpI@EXuJbkdC;C!uFmKE)8p)_*$tK<|AFb3>}TJ8
zJZHT%x3fC!43_^Sn3>O)gS(pqigd=qP_~^Go*INE|G{WJXAYGq<6$&G886&_($>0G
z+G6Gi+ej_^i?_g}+4k`FSHP0d*4Q`06_uQ)cfML5-|osFXMrJvV$LmlI~UQRQ*qhd
z9{YzzAW_`|g?A@l_?lk0Bx>V@vkqQuoCvMH2T19108K66Ow|R7_`59t!=>NQg$8?=
z>aD@v4i7|XSmNC8o3za9B;~i5(=&+-x_4(T#(WzHuBoS->8{wNKNa3dT4?)yn2PUn
zmic{=-WYzTu;GI^zfoFn<&MAfc}X>GeMxve<UC2t(KxgCcdl@FT@jsmStdyInOGgY
zZ;54;u^U_Qc?a)I%@ND%?^)U{7|Y(YJY>|9%ajh;u#YWXEUY`7WgghgYEGsxja#lv
zmhS}9ysxs-&oXBpJTijj>%l1DJ$TKosfg--i%b&x;cloNegtra>vn?XvLBQ)XFP1C
zYM|q#B*I#`7vuDAy1r>N{-k<h?m#DaTrohxv@xjS4&?n+{<z4wZCTu#F@CW*B$*HL
zIrFWy;U+cS*T#F<v8dC3O0}G?ob|;OUbs)=uB}I8EkAD!cfq%ZlTg_?4MW2hqLe#*
z&M0c&OSBn=_E*NbL{0n%ZX^GuEu^sCA7hI8BFTaK<$G-=jbmQ;z1x>w4;)Usiiz11
zzthv!=<3#~m+6l8E_!jXlSXg5OsMWlee>hU(K?z{DNmx3ft$$3@3N5BcAiBHT1}(0
z6PQG}9w}Wv$2OMi6*>!BncwwI%u;I$vwoyr<5_r^nd|?@{LXRrq(mL-+O?IbrI)b^
z%|>=9;t|iO$z$qVJw(fxz|Mnr5tjU-UKQM_nf`+|%`(O5aR#VUY@@r;gYdS&0^<WG
zA*iMohR9FDh>M&f7vzbyJ#J8*?TJufDq6BMpgq$Q{r}m*<q6OJi9tBm?hI==efaCl
z;~hX{MBZ8rwOPLSZQu&CY;*31`a{wQeK?1SvjJWQa0Zkz!lo2b*FPUR*LNJ`x+0M!
zJq~tmesDA3*WQHllzZ=;@Z_Z@6!_0s>{&<8%gsq6)ExKuwXCgjEG;aYj-CQ}xZL@V
zu5LFMDD60@r>>^<KsiW9Nr;NQrm^pK4s>qOXg2V}R`%50zs9`ShN-x?Gusv8Sp6z>
zy1=gir~b2K;U1mrC7(<0+qZ~$U2A7=HZ5kQ?OiOlcLXzOKEl#_>e<QS7Itpx9u|<$
z`)mZy_5|M_j)N<E^Uj?iG@t9BdE$5k%$$f{Lv%6Sc`|fEhT*5~6z&c+gCTcATQusx
z-q{$%LjvJA)&N7TC*jHD6)@Zqh%K2s+m~v9u-8UNdo}<TMrybb-~pSDqcL~M3LNLG
z*?$jql3l}8wEFqsP^T$M)O|6*ihCrV+$1*52J$cE@O)M@??W2Gqn-CwUn<~@v>Rk*
z?4=EUujz)$W_moMAFNMo!~*#OQup%4ilpgC3g*6#SPK{|iK58S_XQ8XVQ4PlJnJca
z;XD6fb^ZH4G&e{N{=Q+9^Q(j^HoX<UT6hY2N9Kq=CrfGu7xfYHo|{lQ&z<_to+5tt
z8^Cfu#IiFE*J_kkKe1Hs(qcyr&ttmv?(D&$A<S*N2`l3}25T2S;Z<p3R@A_Z{z;tO
z*J~_7WB5Jjm?l1r9*<eGRIs*F9iLjd$%tpeJa^2+uL-*F+&BjPCY$2&8x1tgcftq$
znVK3RkIWyQ*jo^e@U6VpGj<j>E?SAw1pyduyAfprWiaof1hj(YBYvqBY|aLtOy3Gk
zNik?B)rYx{1Ja{8(`VdT?C~E8-IR8^=Ig|nwQJE3GY@B5SL6IQ?kZ28fu~Y26y<)O
z4jq@oGu07T#w3t@Ae!Ds1Y&gTT)gg3#itx&xcXh8re|%!>p8E4pBI&J&?%rMv|$V;
zy`6~b+&5)@%}<<Lu!G`f1PeQZBk8g^=NtZ!W25dxi(|s_#c0E&77_{5g_1%3Ldn+@
zF~4h@aA(O~F@AQ782CG1RCp@K^7h^pYy1kC+2kbl)u3A}-SC)o-fU#va(~%|`U0lB
zR0?6S{jsH@FB(VdK(5soN^zPnJ8uC^_XT*KF9~@)GYmRqh@Zae5fkHrhldD-D|nA8
z&;x9_6E4ROz@*)i(dzDqW6G-#?GTDxuYc1xo(oQlmxE?C=gQ2=K<-D*i808)SI#_Z
zv*fOBeeOX~UIuy2sGDq}fqjMsShi3OeY&?|Ti`0(ExStEJ;l^EaSP|v+aR!c9##zR
zjc5sJv{*Z0QlCtEWVHkpb9?AzM~8*ey8D7u&R|R~*+zHQ^@G33Xl(J=Ok<Mk==F>)
zK|`~IQu@y%znzT4hzsI|{?dr|@uW{jE>T<KYB6s@gK)9+n5g@-FJ&4|6O_VrX<zzE
zRzFLYyX0$FUExz!5V%M<=<Fn}di{o->`7xM<A$6q81<X2-m1^+J~B2pIf`8vHVL0K
zG$7A?qV0Zjp~pEnGJ0ci<mU{e&Qn65oi6?iP=jPvA50Ek0pFv%qij0?1zkb7SjSo8
zH4E{pM-R_a%us0{g$H_zk@49P`*l_$d-+HVniYm#OH!e*EgI2x<1vQm^ZXcRZ=Dq2
z_cxagI>(}?IugrfX9B%8BRnStckQ;(;LL39n^}(1sJS@4#u*A)i3rcifXkbsRI8}U
znZJFo#jih#-rS@c-=*N#;ZD=7&9L&&U`+bY7KfY+An)Uj7Z<u|4?iylul-5dehb;e
z+9$*Y<Wk=0+vLpmzu|HRMf>ze<l5{<jXqU;Hor)4>9?EmzI#xw2Sb@=lRArVY7?ag
z-DRqu;zUE`BO;mPv%^ClvtOLeo3DI|c`R;WBXW<j!}r?Q^I$y|=4FRnCr2Ym#sYue
zI>5QNHG1{&!a9B~ns<@+!9sqLx7QeW@>xC~vmv9OEppe!BZPk+f4=mCVIM^dbm1Mz
zog4AMd?`Y1EJv2<1jH`Whqjjw(r=|8R@i}Wb3M>k+5$oU^3mwK9UoWQV(X$17<@8-
z$=ntamU9Mh(p*&VUSVCEJl5;9(4|vT@cdLBZ0dP#`cI7@^Ee!bcT`g!9V_(qmgU~~
zalG>)1KsljVW-wc37?Fy_-h?)dTf9W1q-}yG==}Xu{iHrOCDuEX>-4i^kB|ATKK$z
zJi1g^`G6vtwcm{{^xAPI`}CKZL#Gxne}{|IAhnWNpK268j(Aq1_G7E)yHc8&{BUE+
z<}obtoC(t#Sypq#<G0vjGh2-IE?_Hq?PLyaku1s~fqm6J$XbfkU^?Fs*5&+fbRUEd
z$<tsu)(P&M<>I8`f#LGr5V~EFdU_si-$;RFq%Y5ik41R(e54KFJsaCp$SpF)+kd<#
z!aY%^??@qkg%myxiNF%84BW|EjW>OAFts&^b1XLD!_hR(16IVGlRME6vKY4RF^~@q
zL5O}lIy*h^j`szO9e5A0G=#J06X8&2gp-_Wdu3QF`MYhx=z%)uN(mK?^l+YvVkX{i
zizUy=J%ZugM0^gJi^2^Obi#2gzK%47!do4jy3aX1>*Yu`PYqkE_6rd&rjeShA(GeF
zSUNcIj^Ba()sJ6&7QI{dP#Lb*NT_6x)x0X&`>H?lUC~c`nD0fF-+qeQq{ov<Y?P4s
z_=EU;;cWK!>2EP_;<K9krlV}_@GN$H^#vv$?8+#zlF4fDPT^*LZ&eF`!{(*9yW0>2
zLOdP}3BltAcgz_#0d61Wpm?bn=J4Ln%hirh*Nwo@qth@(asgglQi0Qz6nN>T;OK7?
zEZL=tt#3Rz<17T{Mux+466bvA7hp(%51#AgVWG4;8r8Ppy;KKva!<`_g-*)1iNd5J
zSqxurh0G2lLzLZ$l1lELKCOcbjyyBW9Y_iX`eE?l5x6^a4yyj<LUMmLM*doi#)!RS
zGAagbBIwujcsR)2q*}FCRIu9*$D*w1=Cdbcou&Z=wH4$No=$p>L<9J>?pa#BkZjUP
z4xYW(&WXDO&541c-p;3D`KPd&J<IpfM9l?EX@NZ3{N)6Dd3u?p<p$1-Zw(d)uACrF
zj!a<Do9%h8TFT<9pD{B!9LoZYUy6=f7%O)=$9jgS;p1E#Bz<(i%`1x#bypo<Iu~G`
z?OcS_I^gvO9aMkT#N>^dn0~|w4t~BEr;v#|#eUG7?TV$^oUP-Yk0p!NVDqfOxa2_m
zc{~wOgA<^-G8F&a$mD1FZAfewkNB;V;g`7y)-CI?Y%b@3=gGi5kTbUJgW#zW%&#Y3
zsr9BTRvIpb-{<GF`5>R|d23<CKTWizEW}YyK4+0!&G}!w;TEb&bDYP6T{NIgoFRXn
z&mSvZa2|GxHDX4*AjdNPd>wV1#N(bg7$VJozdiZN1zXgp{}$go{UmhBsbc6R6{Jq^
z6v|wW(7uenv_iq1KKdM~*`#Yl=FV3wmHlICS}u(zXDu%wS#h3VpE-{8bmWS!+e6um
zm~a+&>%I6f>_3)ruRl9_?KrcmI>;g}9$`aw@H5@_8HmVOjM;zXVdZRyIyGl_ZSX^S
zel+*L&BDXGvpLIClluo^us^{SE~eJ_b9XFcCi3}_q6H4_9R@4T(F~j!#@~xgFw@uu
zwL62*cqtS1o_0_YJTZFWGFW?OVRU2({ytlWuBteEx;BOT-eaJdl#i)(X?Qgv6`7qn
zxKUaJl_~|8-|dh4rn$Jgf$-;&E!6jJ!Nc|8*gs|~R`ATaE@!H`7AD}O%@%~-dnTBN
zo~PQ=L&W=ie~bS<aNgQ<LeuAAcx2y2(J=$D<BJsC`aP1FHmV_q?+;de?oEb+rm&&1
zfpkzO*V4LoeYL~0Y|De^j?`E;Y_GAdDi!+7N?}ib91xTrsIxAeSXOQ|No)@Y7d5XY
zh(%!oYEIU<u~OAkW|CUSzLeZ&cYBqx@F-KPTD=C(Z;wOPqgB|}5Q0&hd+X(Afg5jD
zVzOr}4A&DP>Q-XeY0mFk9}5lL4OsR#2iq5Ot{ityG%KcJP);J&97@Hyzj4s7PQeLl
zV`PO{V`O<eW<SZo5biFTSdxRl1GbRt(1cQUKkU7*9zVaD^Zt|(KmRO7URefix^pho
z^>iG25rp;fN!UN4l=AEA$kZhma<7aq+l>2U6QU8uI;s2324el<>8EjTjB8l{jmgb4
zQ)wJLjvJyra}s=&I1i9J_q2}mN9<QcH0??!m+g*pSF=c<&JAScu0&tX%Hq?(FT%U$
zviO>MQOK>Rq)#-5xu;*G6+WiS^>P8RKJ(esT^Zu{!dmuHQS$8X#>uSr-BjlF@2<El
zupb*g{R4B;PGXWPYuTrL``FuYnuxd_g%NXBLUzhZsLoyvvkl(3^KK9<dB0xwk`b=#
zSPw%0uN{l<$RHKU&q7dtCKSy&iC~`su%>AaMBQ9C{ER|Bqj{)MUw|*X3)w$=0d(b-
zLiu_;eky0-+^$l1w<f{4wHTi&^|8fx6BJXl@k>@3Yd;M|WtTVl@oPqK|53a*WQ8Xg
zfrx*!5x+VAOZJH^cH~aL_^2HiJSrBYN^g08(G^>#Zl@<Q>d@G;m|j0L$9{>CSl1Z=
zk00i!xq6CBoF)s>HVYv4@~=pfji3_qjs7k>AbzWph0FF@(*3)b+Lp%B_v%f;v5DPu
zPnk0Vq()K8uOCABk>Qq4c}Ku=>;g7YZ#vKU8VQ{;HR6JG_Dn18wW#KElO0TQ6_tGp
z*z)mrSz<4#vz4RjSdnTx7HnF9$GYYy(KbNbY|fagiiO2=C%jTff;9Jz_ToMgJKpWP
zZSIA8eQYq(eJv7$Z7^xlUib%?!{X^myx?q~%9`1DrC|iyvz$G@B^MP*YhW0+4#t}@
z@Z<b)Y?sN$#X+2_-eC)~g{zUOk_h?O2^cgk2aZ1KsQ<?A^OCz@JpU!BDU8K>t4Mg9
zU55eO|7me22BS+xac4;(Zg=zf>^&LY#mo{e_EUu3EhS8}-A(@C)5&S}WE}Wx36~Wj
zWe$*p*URpjrcdKA!)`E&`si4Ap61^2&ZYES_5>*!jU)?YNhFUfs;NA0z-s@SCfv8_
z&89v!Wm`Xau#y?aSy8T<rEkf|8gu)in&&x(nBBnxOfNA<)P51i)Dt$c$_2;Siz_b~
z1=lbkG>P5m9m<R>BH+;MfZJ+1$e*+TI;N?3GAtH5<rd;;dpvGfEkdkoGU{FUYwe$c
zzT9c9J2x0ZRh>~cVI~ZlEU?L82|gH1!1nTNto6)7k6$?IJttvKo;jW#R7YpoEDYZC
zm7+c=A*9O<TQ)2M|48DGRu+Eg#G$`YBrd$(gXlw}arSCFey<!0+qYgYd1!|v|Ar%W
zP63Yh-i^g3W;FAEeYV(G42zqBI^3YYSCe?o;H=Q^YA_<a-AFN}FD$m%V5NStpyaYp
z+`T`Dx>ig^Q=B|F;MlT!dA%5*t0IK#X%%ke4aeaVx2b&oQE_9dz94H7!PdLmP-xwD
z=Ii{#a(KTZ6uo~<H7h6+<CDiyiH$u!UuB9eZ*(~4vx=3!ln`EPq>FY9VeInUGInpN
zEqm}r@@&=B4LpMBjy-D^V8(8D^sfzpj<PPCKM%)tEq@%#TnCSG-XZO6g1M^8@nr9Q
zm_3Sx-n%3W;4Z73Exxe4YygRyG3Zwt3PsIm^e=HkxWiIDW6ni9TLdZR!>H<91ION5
zFlx~z{A<~ZJ(pwg^~f3=UA7691zWIbw-b_iZ-|_@W^xm-sbdec4dQS}a}T8ErlF|Y
z8M19D7}u!^r7GUP9r%m7WO+yX5Fp<$65->Y@$T3f(d0At$hL1GIgvXX`d$-lT`PrB
zer_--F(uPPf8j7kJGVd77nX7M??u@r%Hr9=W9$_@9`Zsk&#n~Zziy-_D~U;%tf-l%
z7D4|x^kxM!{}b~bc(AGQzl0xBLs;ufch<f{L9|NG=b651mJ%C!i+7f3v1{gC%$3i(
z8oRjn&nW>14G4CfF1VSw23liQAh=@|dL82D&lT%&W|SVnU#~^r749`Ii9y{&0fU^D
zA$NBizRy{~`>A}6Wm<?MueU>I<1AS1TaLHm<k43}656uf*v03Fs&|heDmxYvuW@$i
z3`_i$5Rjao1lwSDs4l4IF6=@)zOfAZmhQ&J#%Xvi?+DevL}YYuK7S?e4JW6db?_2A
z5ZWld-3}GbX=oPHaPoXLHH};iouEL{Pxwurc}J^gaJ-;pSzJ@Qb`{D)Pm{i`KD|od
zjK$~kDDK96^0!M6qz@H{703S*rBbJoZM`E~|ML(T+9^R!+kz%`9~HwAuG6m##Vo7u
zZKl1dS-gGBkwTTT*n=}IOk!$=5NQ>{7A4PM5{;|bRx4ZX@#fE7oCB0Gw3fyChoE7{
zHe9px$JQ;}@ifL8ffk%)#GfOtcKG4y99O7kt0L|AVk{1dz_bON9c!M0n+gGVo$tcg
zu)DdpM-N{|48Yn}O{D&eMC*?6s5&tb&$ea3!PXy-Lu?RkSp=_jrI<Uf6khtFXn$#k
zo+Ea6Xfz8?<sGnp|7MKsU5d1{xv)_gjW^$d;55Df*_%h=W2FO5COoFPuzU1%MF#D0
zNkz$XV;Gw8@A=7KxXyY`8E0B3)Smy3uO6iMb$t={K9y(I!;#~*iyZCWQ&e9mSe4$T
z_?45P5L+t5?Uu#!o?PlP|B>Z#vp7+=?7LufzL2eF_KYT$@%*kEQ@D~rp}Uz-vadlH
zn<GzwFP?}8=1gX5CzY^!Zxb!+;}V&5Kq{-3y2q?KRN3!+_RM;o{Mn?4g@}6;jK18N
zci>ejoCCFyKWsM_u=4NOky#kYIkKG}6Y%T33yN&FVetN4@YL9Yh4B**)3XfQdUDXc
zOcf~uBT>~W8>6oOA?;NYxnDd4^F~g?ld7d?{OAIY5Ji5y{YDx~_oCy&R_^kd4}%9W
z=x|zsS#C!$t|J$X{PVdfa{xbYj6r*mDHMwtt=Kz<vuUg`^ypaJ=D*uMJQ^A^_mjnq
z2KxC^2^ab;MRr;|eIH?qmfHUKzSR+$ul5Npf4`IRwobZV$@3Mb+~{SR8)eI0rr}FU
zNo^$e2kL*Ii2fd;#BDv+Qc_GG|Jz<uI4qia^wI_Wfmw9@@}-(its+)$@mz2;juw0$
zm5ML7=ZHzO$FRr06wf;7US)&l7cz$cDe+wVN0#ick!i;lv&+U=Y?=@b>m^<|b7wXG
zyV4<)Sm54e0j|5F5#N@9XSpHpYIDYctaP|)2Vup4Fwl^3*zCIu%cm#8JZBRUVp1S?
zc`&~h&qIT9D00>KPIY=9E}e45YnY<-)C{D&_raT-Vr1NQg^yZ3{@b>LhU?nnaNh!~
zEsj9IwS}m7p@+5s6EI6{J{l&E#9Cf(ZnNk9w8&Cu`MKbbO+NPg*#n=;3*dY=lXEDi
z<3gS@7IVhuU%6+3?Bn+|>d`YQe$2VF7j$u9F40$hEehOM!#x!jsr=YUF?Nz8R<|7>
ziHEIXa?S=Osr8#~84jljJoA3N-*LhB?PbbP+AZcvoE9FO)MVf9Toif?_J{*?8`;0e
z-z<28E0Z|(vPR<O7|XnC%3^fK0Lyf--&xh2J*;HGMX~1cK(=0U3yuXi!s(Yk*17D&
zMvY_)h~9-sa%(Z$nRBC#<|ED26KeS}DBqod<ts8FVLJ<vuBlKPxe(9y&4kU6F|ap3
zg7<N8crwTUSwAKt`rQl^zkJPkLryT)S%7o0`UwBK9&Zj0$Da}XabvR-MsRQU$SwO}
z`&0$xMfupqebX(2<dNu<j*n8i(Y$mduIHrV&+2P5D@z$SCWm6f^j!FcDd76{ofMpX
zfqut4r827x6zdR!(xGptr?HXD1n!+_vq0E=qMj|vbbU>$rKO)bs!L_SdGRFiNK??b
z`jxz&{h(DxPYJcg3FH;Em9*G-i)`T#*{<5lwtQD4xumm#L}I9UX80c==>BxhP~OJY
z8O~&vV*1mouOgE)ZeoU?AG4307nt&nRcxbwn;7xSSnNF_2G8AQW50A9N>f~L*LW{Z
zPFsh)6ZpAO--&bQit%^aA_R98W5<=vxMjqh<wt{2R+oc$!#AQM-IV9;j>7XcKR<FW
zgtW#8ylB~p_HVB6Ht>Qd5e3T^Z^^yf2Bxd{J!?{Lgg!k?C-v=NAfJsTX?FNJZ!JG_
znq#=BDu&HDfsM1RVaH#Go5x~I-tmYw6okO_;3yd7q(Z)U7k{qczNkxc$(&!W$RdG+
zWLuP;i6p<Oo%H+AC|X|jmn1bLkow(>e-0Mo#@!O4@$KZ8by=)fHJ7AUUlW>oq_F$3
z9Q#?=&fPsJlzFmRd}(-(wEMmlqE*|gm&G5VK8-ho{9g&evou3`kQ7RlL)yg|W42KH
z(6cpL?&UF!#~u{;bw9g2px4>lLDoz=NMswg{t-iGtis0L+~ex81gu{eHoaPao6lAv
z=k+v1X~p5H$7U?&&-c#x8_>LR4elr;pz+g2>^#2)s=IyAA{&BNmAs$dkcorN4zRN-
z!TzT0Fn+TGzR!;F&-oOD>MX=F;{y2I{X^wOeK;dzF;-+|!uIG~{7dykUi=mW<n4u8
zA7gx4;fAHQVThi&8KJ+bDB^T~Z1S-|fpZVlxd|wof0OG)8|bfF8eAW3M!NoRwA>BC
z#?@(*!8`o(rDQq#Yb9b0<H)>up<q(F3`Ga+NP6)>A=iY@n@+ot%CZjOUQSSrrYYz4
z4^Af;<63gD<N0gNo8o}hZ}elvV=DYnEv#y}YpI@MMt1p5tW@&4@O$neYPb_ehGB=9
zLSY3f*E=UlEL5$|ke6X+=8a)RDJC@!g66X&h1y_W%rUvtA5#ZT!A+C(nENpsDRUE{
zz}+179uA0VUJDN=N35H)1)}wONEdjZdrcOkOSj|J+{wIqo{lx)j{KSbB)*;q#5?m0
zgm$T8FK2$Vee9uNn<UJNSBK;;X&k%zoz6T>f(>UaC9~Q1wN?`+6Zc@n4R=^-kAa%w
zRr;$`N#=(?lIiw2-1Ex!5Us;7<Apn=>SUq?3nBK%aV~Wt-;HKq&tw<mj|{<pjlO8E
zRi=WOx5y>_yfCE6kUVxiB-y_P^fL9ng_fnJ*!oTiSz~{amfZlcOR_^ecqoaP8Xcgt
zSGKI#=LtQw*d+w2E);`I2eY6)Cs^Q(-{j#pRrFdtMXXTIqo8qdOd@HY_*C^2EB}@)
zObj@~5>|2c%)5Bux%P49c}j&jj<RRZ-nl_)tru?0Sd3krg$R1-0jn?G+=-QmCpKZ2
zqmT?M)o`rd$$h0s$>^@l!as>3D0gTfN8t!Y`0j%Klvz+N^ha6R3}kSp+lNL0U-_)H
z@w7C~gz{%Z&RUxjxerrHy)iX+K8AI0j%CF*$h!L=IbaPM)+~qR=RK(5y!w;ht?;|X
z6h`G{khv9!YWWzt&)G~T#y+Gk&5NMECk+Q;N1}-Du$qtTrl+UW@NLd&!on`{pEe7{
zQnqM+u0%y^vunoent(UVlHX%<#qYX1EGPZB%{@kUEbTNF3Rbj-n7zK(`}{Y0(7_qC
zQjh65#n8aEW8|v-nY|0SND=o=ipk@qi^;QJlWz1(a`l%MVwc;pkCWmoGp^P$r3tmn
z<mN6hWK1&~ntY!HbayZb=e=T@zXiNT`e2yrOgN9+#rZvX2<RJxr^c&sHSi!-Dy%}@
zC^wvxa)8gWv5<eX2nG__nEOfK^RnZxxU?6=J67|#;A}Lk+JYKeKF10uM%~r%xK*(U
zMVm9IJ^mYo|K`~MpD0Xpl|s_79r)$522rm3Y!o~UQW=NQ@G}XU#x2GKMP<B>k45h#
zMo3sc1+8Y<Fb`aWt4sgq`<h8b`a>YuV!&Ow#^|rZU)wb~*c{<=s)6P>6g7{^r<|sS
z92pqRGK5#{D&+s!Nwv)`w7%;f)m_t~=!}SJQ`70>;I>P!T{9ksnz_qoQ6^1n-XR=p
zD-$oO`jC8HAk|vCimn>Y<ea^MUgm!n>#O_)hdFg(r}B4E;>dT4rt6oOe|`x|e|1wF
zI607Q-T#`caV%r2($m;r837~bg~Ml9A@Xde<Lo<6up69*WRs5W9WmS`>xy@&0qD$_
zkFFQ%V6`g-hQ^%Zd+QhsDFNlQ4i|srV{#4OCoRuHKrjc(Je~==05kXpj>1jPt+2n5
ziqo~3+}~-Bk~;33Ul@VF<7)8s*od*Rzj^M%04W+NXu7`u;uAN_=liVZXB^?PLLKM-
zRq*fJ2JQi#gS>s}u)iO}eLs_7=FVp>=RD}O+Bk&Veo7URuSx%+22Muk(ei*23RyE0
zua%8a^`M#axu;n)CpmD2i36<GX_4m<DMUX>u6CB>XMpE_%t&>#NU{5<yf%RH=M{@u
zM_!9_+RPdE>GAWYieOz8V$oI_L}ypsq@smtHQ!^WiH-xWF)!}>%ZgmXJFc(UnNn}z
z(1N{8?^pk`PaJ&l$lD(cxv^N&#~<45Rv6MR5&A2w@kJ{YbNTmEWx_Jde;$R4mfYvK
zY6Y(PZNqqjbf`(Lg5;Y0*ch*eA(K>4Khy+ficvU#?KlvhjdR;%(b*%Q-ywIrdA0(d
zo4q+}avc&!9w2cb_W<%4>%R@h$tGk2sVMWa+<zg^-}jT+b;5D%l_lp+>A^`|6~?vP
zw<k9V>-J8?k|kAiHjG~*;@W8Z*U1Q&ag4lupVEFl`yW%oGd;KaqF>V=8hT#=or`h=
zt0CWLqv0N!_e~bJhP)FFuRKSW4jmHjMINmw>R2qinD~_f<5DR1#9^`+?@Jd7ErqbD
z2L$IMJDK{oc*=Z`C4B1PEI_%%bS^xbDgUt(8|-JWkC9)5K3-}}ms;5#e$Ka^s>xQ}
zR2J{$PGtk%b3fGSB;;CgKl1|t1%vEBkD|G2APGZd9MI<`pFel+<!;R_knQF<Cq7FG
z9Ab^>3Uly+@7`4B#zLi-->dc|VMI<QpZj}3)|z|$)@Nbu7+17(*kDY(45Zc1(ATkl
zY02VM+*!UJ6F6HZbgKbOl~>~@_r#^I^T$5NDOi3(2E#dLt*kNvgV$#x?wvCRdk3Q8
zUIN;FC_#JQb8?%s0>91LC|Oe#e{?nJ*M4JUsCZ)g@<TL1$A<3x8%}b*e}wN<6DcaN
zgLH4y(6do~ROGeKB7E9Nl;qqJri7G{hOM&DD^e0sg$|Sv@PtZ&Jm`ui@AY_`q6#NT
z!JPNzE_Ci<^)ZFw$0<6NItKGtX1Ki=ap)|Ys$b9UtXjx&dA7RZRyvE838*gH_Cd5h
zu(PKB!#MmaOMunEsqj6%71GVY7&t5(0S!JF9Wo1VxQ9pY+a^4lHXE+?iC8%(0tbD3
zv8~1jKa)1&u!0F5F80M684pZcJ^(wOnxMWS8B5QpBVhDYI8RbV?V2&L{q&8@Lan(g
zLW}$OMj-dzej0qw8LxyV6gt-zYf4=ZFv1557-xT3^S$fRC|sB~8ZS0*PTGONh|Jc8
z#Q;D2^U}re+XJAil}wGcKICL<kJi~S<hIC_OlF;<)CJk3+u#JXKf@4xJcEQK^U*fp
z18o`jmkz1^rQc4@!Zm)DSoqC|Zu2}|T9PvQRC9*fef^rwZ-Y^Lp^k}eXGr%|6jMGi
zs>W^6Tlzj~y<j}NO04<Hxj=g}nTe^ZxHM@FbI8pTZdo}}=GaUY`YM9mQr*UuPdd*6
ze=oyU<tg}Qvl4Y?KHSIShX*oAux>NOoKMp++-N&eF3rcdI}XsvUyQ;b2{7%a55Ir%
z=;YqxOV1PdZfP!@!xC}*ohjdAEk|K?6l!&5pu{nq&kF`Z`@cER)XG4CToNSvrNUtT
zYRp&N1Jcq)v*3ok=QJ=V`6F3QSAqq9|N7Mi<F(%u=&1xjj-Tt>Z~3F5$R1<V_tPl(
z;aJx{67eQJ)Eze%*(Y|8w|yp=%qZpO(NVN_j1#`}<1-6>&#|i=Na-_G5Ezz2=Qf-X
zq_=lc;@WN@={e7XN)Ny`85dffr;Lq8mE!x|a!f*TP)%l!j96yAjih_j1v%bL*PRr`
zRBt+vk~=8(Y$D54>cxJ~%VCkLJs7KbEVe%iBE5!I;pw93Vst1~r_XX?Cd*=A<un}$
zdD$>m5682Q{uq?4g}(O}Lc>2CY0C<rq~y$ZbCb~6GZ#KQOBOav7W0}?_<c7G8e?5y
zczz<bHdmpW-!BZzX5i9NT_oI$L!Yjl7%-GGl)sllvSA1+Q+OvLEE4uRhGMT(f3Ub>
zs&(eP^$sWa2Pa_2oLP7fIsrv9d*}ntZ)LixVp-uV2=eKeanzDDx?c+GL_G|>GZy2N
zv+;I92$p$Iqa9x*aB}1~;qt&IG&47nQo0*y)TJF{(W6X@_-i?q`B0c|row`liYTEJ
zZ?^l88moverJ^N$EuFu7<!nzSj2&3ZHg+Br?%Um<!=9JKh|1HVvckBU_TVD+342AE
zbJu9YopR>;aEf?+=|s^bnX^Og-)Fuf|5^qPQebo@Mg0D_L0DBgjMs0rLgRuPrVjH(
zCiiNXTNdJ=R3z4!u0>I`J0`zMz|AR^=-9jiZM^f~pB>G)-EPQ!Vb1$I0{+-!K;gV0
zD18&AFLcJ1XztK6RK??K{F&<KA4)DC4k@KQNLi?a=w7NQ9V-K0#f6B!A|P<kFdTjA
zk3WfK(2YvRf^;)%zvl_{eid~80l%J{SHl?<6~wOIOIEU}*t1OoqfZaSm|c$K%?SqY
zg;B_I83T={BdFe@oqr}v1*s$b=|iYIJjzUw*HB1_m!#Ri+nPds+yLfkZ;5iNMCx8U
znF7;WYg`@Q3x+yn6j8k0GVj0(8vap-&99MTa@*2`lAUW<dBRR1-hVo&rH*8V+;j3l
z)`gX-nX|y_K4jhRl*O)FB`mY*t|(DhX4z$zDC*w7!3q)!Siq7c@Hn4>%ZrjQCMu8f
zGDA^Zv=G{nu?TGp#n&Znc>irRrY=c9D4)$A57`PO#X>xDbLKnhSnN*ZvzCt$@a&#~
z;PP-3>!o0AzpwP(Pz%!itgxOtXkz_mLgjEG-YH1nOZ!;(dalCT5uPY~-cBF?-lA>g
z+-u(4L|-!xli?I?*sMB9p~W2(b!#PQm&jt(bHb=yHgNeFM^BG`q3@2@NV>k8dct*3
zX{kbzV``}@d<<6I+)9s&>*+!8dx6GJ6jk5lQJv#pXzYujsfj1WR#_MFahBi?jP;h+
zc<)uOqM+vWypy8kqOCj+a7|b+D1;*R^cQ=ayx4^^2ZdWB<k;S+3t3~RJBy$NoIBAX
zMxT2j8m_-DNT-C5bNmom`!k<~T?%2bqf(hC&-`{juV730J$`?RH;g4m!R3AtpZQG2
z&21;~?CUnTe&LSN&s#C^m^-SB64B_f9YJ^8aVT*<RFl%M?rj!smd2p<#V*tiS&Ca$
z+fY6zm*-b*(tGbNiu>))yD2KzRu=&0V}Ge3ARi}1UA*VHpe@rZF=4wS<U-Xrr&1fc
zA4Vhd;~Qe{89kYwjOFWo(T?k(czk>~sp>GwON_@kk@F9;HQ;6^g~j2Cn7$*On(j@&
z*EQ1-q19L;<2efw?bF1!mYiR6V;i}A;Mb3mm2@G;-Li5;wm5j`SK-$6G5q;@Gt;s4
z64F-aQSj_2y8P}DY4u+&o*P!hwpsRKos)v8{BZz19akwtr~3*YH;-k>w+w_jMQv8U
zzJyBL&xp~hchs0WEnpS%ELm6C4YuG{IcvLJ%<erN&%S=)GX|Z#IK9RR-kpn3`I39K
z8`h)ic{XPqM&rBVc6|Syfw#YO@#f1JT#5}sw|xk9l_jF=9iJZ?Xrpvz0nRO+3GeIt
zXH7N2pJ}d`+~N(rQFiExOn~Sw1-oy)kiRe;*DBURC;Agr>S(~=coLfUv%I7mcPWh9
z42>Ek?B(AN@uo5P_n!$G{*yF@)za{`T<Ta-LhJtYLa2!mwB;|+E=yS)ua%<=xohO{
z>mf<6DWgwE6;ZH}XT}$Qqr{V4RHh)q=RgJ2TZ%g=6VqvVST617T<*$Qr&+`1gS1h_
zfvm4LRQLDXMdMd&rh*xMl)7*jtEfFKN@!))oEfyA^$qXGGzL4+u~%o9SMyHt9ec{s
zQO2EhZTl;RY&*c1;$r$Z)S0;k?x=C?IV5cSCcz%cq@#83MwoH8p3A6pXuP=|9Va|6
zbx%H$n^n=sx!tqHJ^1)C4`qCQ)wnhVFQhLZo9{IO9Xv3O?=bq#n~B=fN1@QK5O(T^
zaAjU9tgq}s)HHvz@xJQE$xXDT$q~<zym0CIC^V_o(&KyiaOu}VBiCf0DDN4ybt^zO
zTm_cX+NnJ#o3e9~VgJPo@dKjaJ5nTl!(Bqk@uyU9gFiF>>V@&@wbV9O9_?B$Nb>1r
zlICoCMdc@Sf4n+n@Sd^x-ba>~B+7-W<4%+A;xe*OpFn43EvL0xpHt*8J7y}MB|QCk
z$8uv;67xIQO|nOHnCFLV7O>uiHfm^zSJR4FSoJcpTb{{2<XmS1Ph~QltUF?RdZQ2;
z6-iGyQ?E}{q9D^-lApEC3PG1vv%Rfnn4U?q_}XO~TzWR3lDjyuc`tW1Ch-iUEi4tc
z@p(}ao*ob&Z5D*ijiZrVzZ`K|PS_pqjiS58*tC!5dDHlwDn<)&A9!xz_<a18o`Nt9
zUo7dEfp+d6&gDDc$Lp<-Tkw|DvJ{~dyc7p+>%;EMSX`<|g>TUp3cou50b}NYoiIS3
zV(y5Y=ZbP4CH$GdT|X+a@X8qhC8sQucLzhk#1E?bxeHxR8y<rSDX>bwuzz0Ek^hcz
z`|cy_x-%Abx=V#q^CYmm`WC&I)Q=S0jKp&ir|9|~e!t(<Ci<mIGPQLpSZ(wrG4e~T
z(B)DgxbD9xgk5tLC5pT%C2l)eqzCq)$ObbyxK)`I@%(-`JIS_WZ)fF0Pt@p5OBZ~f
z4;9*vucB<37owR-D(T2)F>|fm;;qC?(Ieg+l1mbCe54s#R=Q$;$VzOIEr6SP0t)lp
zK%Lo`e1_}AyyBto(+gxh6q_~TAn<&{f@2w2^qqH^o-M+ORVv&i>4(#QtWdsWCQLSa
zAp3|1WXAMGU=3&aHcrB;I5Ye#UkZ(%LtvcULO%u=qp4*KvcENw{a8b^PgR0)`&wu(
zT!EAKn#f*J5tZx!g<gG6oeeouxcm}XJ?f#~BQoi@)exxJdg9Ch1vp9-kztH9zji1K
zeM*OmI+J*=@ADV>R4N74$gxbq(#LXkNEjLRa$x=CtjPJezfku0KhgBpCo%u*VX<D(
zkTX8Fi_(TK1>fENHEY+li^?-b)5Mm2?8~S;vd(I-EYKat{2rWQd3RT_ml_kqzdKg5
zf&+1)PH<nA`e_XF*t<u#wM3OZ&VN<Y!k<fz{cy*Vi`$Xq?}2@X(lK>YK2mjyu(ekZ
z3^S6^^CFA)aNV%;dntA=_duqYh``Wk=*Kg3GLD;Z#*t_A^EP7RH4o&;E=I1&NZw;k
zL4x}n)Loi@&HDZEI7Jl;jCIiNJq&j*Y4Wa=1v>l3l8NLinnXkKgU`8-8Ua=NqVeHx
z37uZ74lC_EXedhJRKGY(y{!cApDKv2ZKi&ueX;$EJDLwlz<%iy(z-gm=66&U?RYwg
z{@dM7yEO-~wVx{}WaMBb@p6EvhPo&QjjR`P_Bl}9;tzuTgF!U=kres<9w<KQvx<d}
z=%LC_+;b>=wETE)xv=v6N0NxQ6<_`y&8#+^V{2{nnRSVorM1~;>e(yJy3S_Nz0V)S
zlqChsb^e)}%mJS)cRt&}G$dLCxjj>uQI9_sUF0({T|OHh9E?@nzF1*o1MmNw@mC=X
z7q0Tmnb8Jht(=9XU@J(i4&@o(VC=cUnfq6!p;y&mJT%yYu6)ks4P1zk7y3h6lh1yS
z^K)bo@0B{}VKm<*Tum~9<TGQOUa=5g8f|fM-vrF*ZGf*2H=_?JLakK>SN{%0X|H^`
zn?DNr2c*!&&qzC;@eb^<`H*beN@W+6pqKN8I=*_)UdCCuLmSBHa2~CVd?&oUT}=bc
zlc+P=9M#kLPQ67H>+e}n*?_YYFZGC!&zT@bJB8{Yx9H{Q4DNgK5ZHjRWPU`MX&7x}
z3m;e0xtRw<?SN9UJd#O2cMN8dTb!9h-s|eZUVWMEq;t&Lz_!NKMv=B0K0ujsxa)La
zv~XBUm8}ww)WrIR^UP!-D?dAd#hlATK}tM4duOAw$sbuI?ijL``&{Jq;XzR`wzl!C
z1b>!y?ux^Rv~ZZ2ZiS_6FfwN@#)j+5;qM%eDgRbrrll`7&9Me1>%i^cOvoB4q94y=
z<X*VKnG&3v-fWKVJ?aQa+l@q?b&BcHN8q)^*m&v|<-hYs-}o3DQm&(0wh2%ft_0Ul
zhiUO4L$qxkfqN2b=t;zO%5_k~ujq9YQtt=ZnG2~3S)}y$i|~m1n!Ku`sW*2<l@HiR
zpPh<Ga{Vsq9dL{GrY8%Jn-7xeh{M$9uPqA=X&0tOnKRd6y5gp>+U(Bb4b<y#3T5@F
z<vEppO!vz=y4PyTUFF|vwAK%?lt|x5UV+-ecPKH5^Eb~(OgUW>)W}_)dmpkS)oAK<
zVhi&QIK>jSXRw!jy=bbBJIm;ri(!c?kjTBTw<R38@4y?Lw+<kA-8{VJ=lm~;OJTy_
z^K(t%a4^Wkmf#@#wXKBZbUv%MkB9NVdDzm#9SDEdB5gt@&eU#!&1ZLvOLIo)W(E8o
zM`s$A)BCmIBuS!CB}s)!iV*6#*GiIvBvg_lb10GunJSfLNg6bY5~UJS&t8Z~=1P&F
zBq5m+GW?SF{=Xl5ayX89>}T(NuXSDLS*z0y|6ad?p99@_-g*>j<U8Y#GE3}Ck;Cx2
zLvi(Jo=aIi4wJU@!Th5$&~TI$M(BuePuyq7KXnWqk2J$1^8`3(!CijCzQXm`QaC-`
z0Dm4xgtmMqC^s!;?HfNqnYIdA7I4o&qZ3>EB$e5e#tZ3n69lQBy14sI6J%N)W3KNP
z)4C1;=II|`t83-4GUE&@mGtH_*#Csc@G=+?IRF((zpz#chm>)W0+>0`&QI?2YkaW~
zUXVo&_Y1^@T5W8t#dxZ8XcWpn1q!Vqdr^Lm9K>_~l*=_>H5T4%_&LNM${AQa^#mFe
zq@nqywWxj}8+U%%jJa(IxYT+nn)7>Y@!l|uT9k&~j}PGA3CnR3&#JEsI*K{29ys+u
zI{IYI#erI*(08#h-mnVAhumorVLuoZ^Q~~w&!IS6`w(V73FY3{6&N$!7#&6}!4EeW
zhGcP<)DUBg%QVNYy~*%%)?#!Scmr~BF2a#@M`7sETcE_*Amy!1F!!<(!g^&4jgiEK
z7N@}Gd>H%ClL<bnl)zPG7&?5cV&?1sxX4I>Lx!ukesH;9a<iTJuP6bn<r|?tsKb*N
zW2p7^%L>og2S7sWmZ-hRojU(p01`i@h(f>4pw`eiLcG%kp<{3!dp$0Nc8a%Bwp$Ey
z$oRrc@}tEPofAZsg^vXZmB#6X!?se%oo7Pp<RY=e%|3B!awbVcv{#tEjo`D;V9tMD
zhWmy@qF-7dY8~Qv(;Mk%a%ML=joyuC4TEqz_nnpT9pu^RF_^Z2zdH_?jn|jvVc*Q<
zsCYF2mzX=^xC(nr`o=R?ybnCFd>yVD#{1?ild<uYJ$5B>|K|FG*yo~&Q)1`hspE_A
zw@M6-&r?BVC*GGn&Aopo64Cq49`vy)gDX!%(dX4<G<+}}eSAfDOs5I96?$UyhdVH0
z?ij3Ess^*Z&V|wqoX0fqAGn`;&rGhz!|k^k;B)01tQsv1CgnRtGc*lh(y0+>I?)8$
z|IHV^*LDdVUrvc8kMtpf^UA`?3VD1Gx`?wVIz_gN?3l$6O=_=v$=a_)3%#RXu=f`4
z!FJ?Is?Ax&G`=QK&lr2Ur#P1qhi)SyvlD`CWhtLW8;kmVEuxs8sWhlLmh_#yu~{<+
zj|T6+byq|2jITdR8++n`X`Asl&rK`mN21w-a11-?jW!X7(cE(;POF@O4gFnEELw$c
ztrz2bo{f%4ScPX)b#Wx~MYpRZ_%MO@9R8T$b?1q=<bw;=jOv4o`WUp2-Hw$!qxz!l
z5H=V7gg}1&UEEd)s$KkhFlZ;voiz~mr9^`}&+LZ%_QImBVc1<=4Vte9;l^8A;nWIW
z7<<egK061%^AT4d{^}U;)Nd5QtTvFkCeO}{`U|E;d3>f~D0og0LuBC`xU=^>)E0*^
zi7&dWM*XnJBxF5n`nQxR<hzo|mk?2{Xej-9e2blwk7l7Gj<a&Vd2H(AXxb@QvboEW
zNkRF%NFhg^f{yD`w(3|>P4!KtwkZnAU4F3-VP&NKZ>s36q%l;U))lT7JHpN_UKqo>
zbr<IsV$*A$p<Nn=Ipd6Q!1pBH?cRtf-#26Z<h5uzCJOt1OT-<~ZWuVy27S_Y;e%h+
zDEu#%FgpR8s<qKsISwVQ-B5Sz7;NF5lC<PWcqHfryo;WIjtRY>bx{vRTb(h{mH%Dk
zW00FT5n~Q%qFvo_2<VW)m?1`3?`47_|G8*(Y9L1OEVJr3Eu5Eg89Z&f;nK1IQ1@94
z0e>4o!L%3xKjgya{h6?`IS&&5R)Ms|UAQ^sJw*PV59>nxLBW5LD0>C(Z9KAspebd7
z-nZw1Tb3HQ^|rI1jp<A-?ylhAq$L<sIS36#o1i>;lOUU81@~S*5;AA4r@+BlpzdDI
zO!v>DnqzZ>j=iaZL2V=}Qrk@3JBIM@fe~fxiiVV&1lDzTHSJb7N1tb?(n#AdRG1ou
z)~)++Z_;wSmYIa7Pp4too2~eQ=VfB_BhlnRAa?U?yl2xcyehvM$!ZDe3jDgdxDS`C
zFvryTP*hS{g_$>l5f&w4PNp@!nY;`iCdQ$aRV7?8utO_UMWurivAALaUe|aJ#zRE-
zdi_?^RNaI}->kuFks+{C>p1+e>Ve)@B0Ob31miXME`Dq!Y!4`e7yM^m=iCUl&v${T
zW*l6Z%pH+~&0yQ%5ioCR6PR`7L%PXuIB3@gl|PTNF;UqpYW){MVc-B(AMOJZDl>%m
zRC`u)X*O#by;*qn(v;wsE#-&C2?l?=1hegYj~^Yv)ZbV_yk3xCr)W!BMK?ejh6?*;
z8dLU^K%soNrBK;qF1(9NquTiwguuvo;2`yiE!%XE3ZYpT)|f{U|Jo|d&hE!6yS?$<
zm<W{q6OTuH`Cc?`7k<|b!EG*%xX{lPPY$2Sug$giFEA2~-v(nquq$f#Mqs(>BFr(_
zi$;T7G4zW&){fnUq60_yZjN^+Hpye}Xf?Fn!ms@`c33lA3kTh|z|YhB;rTdeRGP67
z6CGw^%lpTmo6-zB`8>t6a|R~Je*h={-(Zm$4xZt;&<aWzlyMFmh7Z7<jcM>#XaQB_
zVOaC39FAs3fzhuj=!$WIk3rS2>0}D%b*Q1q#a?mBf5<{i_^$M1j#y&kmx}CHmqeCh
zdW08O_2}MOD=>9Cz$)Jl6Kk%B5%%~O3tn&jv7p6ES!>y5I$(5wNc|`a!v10f_aoxU
zp4y7#UffAB(+VV#`m>^A^Qe972Vv&)vk<YVpU|{u64^Et3zad^=6+@sEa=rXbk5(1
z@x6O-ozWr8vkSn4B@vkU#G7ZMJW(OZ9Pd0$K>5NLoYL-wP35~#f1)kU8xx2958L3D
z02TDO>VbD2YGQ}J0{*#k0!?^cPwt!<@8T)ql>F`ZSfdUsJv8wfe-4?)Ib%ThHyCkO
z2P@rMU|jtHoK~-hp6g99qFM)|JU8MI$I0B0co5#+n2x6&n&BEXe~6ft0@Gv{f!6#_
z;5y6{9`2tDnu_i4s%-%BkUIRn<PLXw7lX_@?ut$r$%u^=;@ca9c(ezHVq?Ll?hI#R
zyNcD-ZwQ*bPnezMd0|kp05aU+ZR=gcYJ<2-Q%##pKb>PTBUXcdbR7LMek4f!N+iiS
zwxFOpOpwSa6_L$vLDN58@X9s=DZaN3+UHr(JMo}U`=>&jy=AU&21b%%{XRac3`Yqi
z7ktipiZR^RczJ~j-$5ndZN&}PF?1as9kmP32du)L^PAAdI0%=>*<$#&#pvB9!Y5zS
za8K$KG+8<xHJgKR&y5|J5@U+a)>f$VU_5^Ryc`_@+|a;No9}>5bDtM?p0@45WJfc!
z*d&E2W!E5l$`Ukvl#DG$E%DgeX1EIjaoYJdh}F3WX;a^UqW5O_RJ#zWs&vuI{0^Ly
z+yTBx&CvSmAagtO6RbmJaE?hC1Z4Su<PAR<*Yp-7d=en~b^+LZ>|<+J)G?PW13^>v
z9c#}p2g!GX@z-4=X4W=R*tu}8pt9PN%=$^Nd@Vz!J*5lWnsi8_T|(%1Zv{c-w<|v0
zzsl^&dYQ!cOXdn+9*auaR9J&aHLE#&g{@YbO+CNYgTuuJafp{5x#x_a3dwb7$UV{H
z&K|;P|D4d>>M({1`*7j;g;*Wri1m3JF!p;Q2JY8HCe3$8O1z6>?Tfok%)+ahrr5ak
z6xuY$<0&m0%$}@;;~sdUGoNiP_8EcG-pS!_r4)239F0b|>OqpT4rjdGg}EzcVZyMF
z@N=6phIQS81r|#9N52W&qxz%i-Cp=S$&k-(Ctzp92)wB(i77=LP;;pPcF*XK5^EOY
zdjCr-xk(i&XGBAFTRG@deS`9+*3cI-6TZy73@>#fK;=gZJg8CSd5Zy9qp!?fOzLIv
zYsX>yeMdH?LXP%aNM!}rOTo_dA5#drCEhb66ju1AaMp7kyL!G7y6XpUALll(Y&crc
zaOX3N@>BvD30uyWHzWHae$W~$78H(!i8F@{BA?0;g0{PcI7#<1wJ8rGb~Y4))Pu2l
zy*v6`*?{Y9;&D&sMwC``N7o5q_{}#Cd#@e9yMGwIkz0qccNgO#kHwh%iSO%-J+N~7
zA>4Tjv9O;u{`q2p+QyujV8gQvi+1w=k6>h@`93clF<_YuK9>0bD>V=D&eS-}Tf*H(
zt@E(E<vPSF?nSF@$KlJ%F&Oh?2hYJ2z$z8aMfN%j3ZVn>%DO-BF-{v#W@uu6cg|7i
zONLgBSTL#RhBr-|+0&&DveA)H6P3h#g6qYhc{bql(F<ghqXluu7Fb-yKy~>QFzdd>
zu6y#_)0<^N(2KZ=rf=7U&``1PYxicBe=3*zdrz^5hmIuCqQ`1nWSE4RrFiL@GB8;d
zCIm&k6I<T6YyQC9g!Y}*r@m<h;Ip+@WSONP3i>8vZn8j<c_fU7j=MYA!yEqSowy%o
z_k`f&DQ<W;Z!6NedFWA@gqs}mc`jlJ=Kb7?7fY6*b;eRO{_4bctd=;TZU)ZuPr-gC
z>~Znf1-K2o(K=Ee<Kvd0kDL@XO!C89y(%apt%3J@ZPDw434U|8#&fTypzfgsXdO5M
z{YNIF)|?Y?f0hMmJgtT7Mdn!Lm;y4<qwsz+_w8<33xy@8U~SS%xRu58LJs$#-$Qfo
z-`5`l<7eQm<tgwQbg;c3jb$&q#2np+!alt>us;4KNQm~DM}(*eA1i{HO4%^@CG;0|
z7R#`fC(}u<&=4x``HEzwP8Uq28Dwksih|6=q8m51gNMaUre2~C4QKR0%J+p3l<zNe
zeTo-L91US*!bE71RtM*dY20mE$lUH1Q)u7;7T_^})D;ZqR#hZ@cC>_k`+_m(=O!#o
zb4G`gkvQYlR-DbxI{_}+F?8>I?veAy&q4F?-;yNE-wXKr;7YX0*p3%ItK+}!`DkL{
zgZE=R@z6ick-A~Ub55T4MsYXpu-t<y?($sgBQ;DO#hp~i%kl9@DJ*s8v%O862QXBL
zcM-2bD&Mg#%=-o_V^pytXex$g4MvfhJO-A^Vfs}ixO?>z7^WGc_(?VNGz`IQW!7j|
zx`#iHRKVKb71jp~1kdaZ(B-0orr#bw$(&oFtg<pN-+dQ;?NJjPb(`6quZ5tNpT@n@
z+M>m8IP+ueYEV!3D@c6aCUW^Rk%G4EU~jh>fx@8MOu|n_<WTfP_&)6<GZ;GnUgc#$
zWl>Irgqo_@ZF#fM(DhL8xZWo+(~c74CLa^jCHFFGnQUQNkQrI74I(X1DUy1;8&juk
z!mm$k@y@cns6K8XT6p^7--RByqHGhU&g9qYD4ze~89g~)CyYOOfOkvca97|=+||7V
zYu<Ctq+Jerzji>U$st&3<B9J|_F^M{7fVR9#6H(asFld~M2{w7W1I*N-r?^{Mk@Gp
z$#C=twnpFWns_C1BtKJ(z<pM4Vdydwe3M%Ur_ajc3(4X5vsWLb`p&}oe%!fj%6AS`
z=fJPv2;3g<nKSL3Agh7Tlt!z--Q!0=Ta|M$S30we<DNoLiydoHl!rVa6O2}fpyYxn
z)c#xoQuf6{RYx#*e%LH@ADBxgL*ziNFkgs|h!Opn5f5&c(plGzw_?-I)2wLc5-^C%
z5=>?u692kp1TKwd*o)D>+2`Xrq~5uiNpv`g?J!r65Ra{}RG1|)96pvle%eftv1{=U
z&)cuv>xz#zAH~PX({M%ZT(lG|N4q5c3@LKNGMlMr6S5Z1U3Wywo(<S%Y=T>+asHO`
z7POl(4HNkcKWDWrDt?N<&cPZux!o8;xkqc)`Ryo5RY04v$tWXmmPV8!R+*UNn#KP?
z+K1VgG?ed1d=*eBR0*XrIXlEu8@ml9kZ~T21h0qIl)r<`*IdzmTM6fd-v!A`YdG`k
z51h=`ho|x%L1ODe_SQZdY7M@N0`zx4tHN4lR(Kf<-d3=b&Ko?ZKOJPonM0SXpU~m9
zl{@3TnZiP6rc`GNE}5F(&~48ukH?8U{x~v)gKt^i?qxu=1+eE~C7bZvm%rP3vJT~I
z%wT5}b6C?S>Z%{l!prQay*~*J@oHogTrH>v-(%LUcLkE|7Frc%3bi^7;-3cs@nGvr
zw2YsPCtxM68}E-pUWTBhEzdy(2H=nP<B@N>P;W#CP8hxeU%Z-)Z2|MKXom})oE(8G
zOIBi}o(L6(O~=JwR^c<dg;+fEAa<`_ingWFxHm-;H4SvI*Ubf&F6Mn)%kjAVGS5jq
z`~fF9_v7fl-|&Rb7>C-_!9%m*IQV=Cyw3OsYyIZp=!8G8%jySgugii!&fr0>bMQIe
z1?$@zplV1bjQM*I#5zmikLPIca5>EKIhX!?=|k{&x{aOXJ0h2*)vU65oY=}gi#ww|
z%Uhqcu-4PdAmCOGNOaDvh*_%#l^(icTSY1Mt5O;w{%~HQt1(DCuQc<090g9+@z8Kl
zAADAvVqIyG%vIVOx|Za!5{(N&S6mV+Stun04M`P6_#9>u>;25lb_6pss}x#d=#Czo
zuk~Y(JFbe~jLmoV;MXP(yx_GG6Tk89=R<Fdvv<M&sum-g&$~H?=3~HO&X&_yjvs7n
zc$RM-)*EV~k=<}Spu{;{xA{BdvL%?qy^(b46)5N{;s`#|>OHH6UFxzJ&-*!NF3jNb
zzW)FgD56iQGkWST!!O6CW6q%~5Y2a-Wxh*Le|{4L=jOrS^W3ArxnfcOF;t$!dEuQk
zQ1$CLEExX@<d#U{>nRx!S6u;GpVmR8xv7YqI|M<S-c|(NzEjcP;s^xwnF^YnH7sKK
zU1s)q5gQX!&6aE3g7Vy8CgCCny62<8tNk!DGAb0-*1u=j9)_&8`jucmR|2$`^XHRz
zDy!XcLDYJ^QTP=#QAp`ScBM<2KCUpNkJ{2?S*$GBcGU~j6)NQHxt%l~O$Sp6KTOKE
z#WiAIjFNN3UpD;t-(-*TXD!FsZEF!tXQNS$3Es<dM%L_vQPcVP;4j~Kn%m=-x0}(k
z>Hsci{|DPA>R^9mb1ZCIhWS0?(XMR-PUu;HF1;IY*=C-<Y4hj39a9YWqK*cE_aG^J
zFcvg5faYuoT<{jLrR*gPk>k&y{Lg&<tAXkTn^4nj5IVls#gm*JUf%kXU43>O<~wwN
zo2nk{jLT(y4zsv>)(r-JXn=zgQeo!$6p&aT1^Erl?5#vLXNGSCiGQ2KB~Na!wYfb)
zwuP;@Wu7Exb}j+W$!l4M+&HG$70i8b(PAIzVsQrVsg_@rWD@`K%e#y(iN74-zCRdc
zUijrc#J{>MGO-C}ldstkeG-JAbD8FnD;JQ$k26A{Ws+beYY!JX6NE19_re3|RoMQ{
zA7}iYjYsZqu34fUMuhXvuh<urYF6RHQ=4&!>U5MFVS`ZXiEWqMuwj%Vs(xI8>Ek?5
zV#jLKp1U7^XgZ<uO=a|2VT6l4RWSdK8gB5OfQGXN;q=Jec$G6g=Smsl+4yP@jwz$c
zI3=uCc>yW=Mx#2<uD9{rPY>gr%Kep`W#0_ZiY?GKP6mrVwZk68TDTDY5^nY@1^ovf
zpk6f?#Pg(ay{kL4RPTiL;TPHSiesR8gTdfe5@@EZ!xDDvf?O9dEE?Vdzqr$|!+buI
zSU#3%UDbh`8!kbn-fF0|c4AF$t_zFopM%V*ZV0+~#JtWn8DyK4D0`bUtBGxffDt+n
z<Y*vd`&fx<%Fi+rekLv-oF$mWB?~W>A7<rUg@XEy)l^i!n5sVvqze{_!nBD_IC|}F
z9AXuZ-H)fBiq%}c(_e%4R;@+%M|1F+s~OrTdE@&ccU<u{5L>SDe#jkjlrS2D9_1@=
z&Fqypp>-;T(`c+H(8G^>zj?Sy1@$kDMQuOMSvp+_(vtJA;HfS~TT5|o<OsB-br|!W
zXCM=Y;sX0y@aN1q(5Wv0)A~EGcNTZ>o_PuSS&w1W5=4#WFYu8b!G4!wu$em_RR8#a
zre7>{tv>@{ngp$*K0~0Y6P$K0f$xot%yFMISVzP{){Y8zRx=h%?d4!(d;w%f{1mPI
z;10bCUzq&IerP+<nZ=*b6kd$#Vb^AqgITFBYl<+2k22*#$&pEd#JVlw?*qbEWzQsW
zZLdghSolsPA>n_~G^$>(HCQCHZ&M=4q|rnJl7!Z2(c+M8qo`L;il{k5kYV%Kvr#ix
z`ZF7Lt!D{y4qME2Hf&}xx8qs<>tI%ElFHh*#;_L-C)u-}0w&dQky+2$&u-ohW2P&1
z^SR+Z7NN0%sSovJSLD|+FO8p~!g=>aHFXxEi8Zo<+msx!_C~{svR9*og$*{soxqo3
zuk;&YKfM#;4xQEFdq;Z2HA^$Z<#U^bUpvKu!Pl>Xk-sz<dL0z3huD#HRilu8N{zH`
zGR`YErTN7>$oJn?`kwDk@?D20=SLd#56Gq&aaTyesGLrw6jNM%DfQ?5ucEovNpanK
zI$+yIn&aP7LjEsmm+qmkpWn!>@C(Ha`a{!Sy&!d+Uv#Ogm!9`_)1QVHr1a$--5uIQ
z?(eHeXY(C$RK86fBO2*l{#&~6x|u>weWt#+c3QmSB~`xpM!Neyl32Zq&R_mYHsgNM
z&z>NrFfWctC`K~N?ieO<FN<|6CbBJ?GnsY7HD=Y2$(|}yuxVPUEPVWPb~C+`oy|MV
zeoc#K|ILqOSdh=e@!>48QJU#LZ5C}+3}bQZspvt9CrkXWmz5u$#a*?M%)w<81AAnq
zPe!t?_(GA_fF6;?;ZtJU>J8$O{qqFvcVhF`5kA5dUoT;6{}91#o{W$;?X@6NLPFPd
zL3r#xhF;zsMsAC&XzH-ZbTaV(X;>uCv8Mf;1)EOgFVg6S`!T9MdyMYQdPvjypC|9P
zH>feQiYhKQ(u>s}NJ;uFrJQ?9%7b1}!pGl4%AM4*`ym}OZ>Qx$x=C)wO>&g2rTm|*
zRQjNuR%Jb=uvzygue*-kT)D<`RJUloZzCmKY@j`f^|U&#or+$!QlCp3Z9Mg!vd4ZP
zH|Za=SgVRjKQCd6($BD0`|4PJ>R*<w`he}5S<HMUeP?&x-)5G_zA-_vfH_}~g~yiH
zSnP+ZOvk;N1s{FMlp<4^%-Ii2Q~f*(Z7gKI4a?cSBmCbND6-@4@>$oH%gkBq$D|%#
zVkvP$D=Mq*o5xOwVLn$mS9Acf%JdWFHd{lPM9PB-g_5P>HF;6OgO^r<N1wTPd%;^F
zN4!c{->goHJEH{2xH_S3QM)kd@f=#_V@!t1v*`K+&dQ&1fW~i#r;fk_<P~2|YWGvA
zUiJe0H?fj(8j9%V(MvQw@;0IOHJb5&f6h`(wDRaj(k-f@b3!|P_54ciwfD)RfV((-
zp3=@=w<$=!m7Z?9K)utNC?>9fPOx(N+xCo1nr_jqmyNW^_dcDid`W?eK2dua@0}j~
zOwVq8qz|2Msa8V*PH$~wR@PtG8rQdM!}B)wm@@!tE=WNB;QnB=X&~HKYXBE{!$L!H
z4EPinLAFOdJM*EPH8?APb4LSfYy8IyXDh)cILWHJAF$!!PngE^681nPm01?WGmDbj
ztT*(&`GKf@kXv(|J1Rf1r)8&EcHtNna_K1>I^Z0uc|S%pD*qt!@(*OcoE$~N7PJYi
z()s2tn<ffssuE;z*jy-hPr~Ij2GqNGIBhA?B=erF)Tzk%nSL{ATl5O5ow|l*I1pXC
zl|qe5S=6)q6wT^iN%JpeQeajIB|r&@``w~P+kenk%d4blc8%I6+$ZZd#q{oMJ;{{Z
zpnzE&R9I6%+0E6|I^`Ofo+zYiH)_ewq>9?k-Jm&tZ&Jp0F>UGJLb5S0=u^fEifMX5
z&Qsr$mE1d;6!?aW4rs#GSIVF{Q68?JRDeU@q~SN;?Y>%V4&p^)!S#YZRCtdAWs|9(
z?QahE<px8(=>kv)SB0-S>TsuEJotsogCD`e;ID-?__Zo?-k1hV4fxLXCIqp%Zv)|q
z)*x6uXcAax#j|MVE39MkAULTP$42N^vNHZ`XukP~1#D<%G5P+Y#hQMsiH#H&R$XMj
za^+d*g?OPR<B-T_|2%QM=>uWkg-O&iC0VfaHz4814EmTjoV@D>lD=y)-57h2=A>*R
z@ueiHt~pNbgRju@z;J4ejv=k%@gyh<<azBq_4hqZD;;jp$yKK*!QdHbJ-WoNKY{Mo
zU7=gtXSm~H5w(Z4P;PYvS=Zg5CYd@aNhl+?D`gbgcaB`oUMB~-LKpg8l0(!BlKlFN
z%-f!lX>>PfEa)JU0lKjKiVj3dDT1ljT<F=Y2jW%0XAw(bf|Dk6p$M3b4P3J{hPEm<
zND3MS!?V0Vp=b$A&J}^wqM5v}tq#?HCPG^FdJxH3fP&v>7^t8JIWNDmY!i2|{3!+B
z9F{|6oJ>Vr^aNf}o(7_Tt4yNxpcz*y!S0h)tVn$b_zb(pM#aiO*7IccIB=|>wfhK5
zx+KGH$e&<R6XS(!P#49A<cmt8?^NjKnFz<;FBEo476=C4R4M4$UNTzC$o<w7%E_^%
zsJ3z{+#5#O7tfJHVid`q-%DF1qew;nI_<fgMWr71Xo6A~Ex2=-%AQ}O+Sxbhi0omy
zI4p-$&&JdErRV6<^9qu>pGex5QpjrH6>5EPj%I&7M^_$SrkH@6B=V~vE$yqc>)3Ph
zG<-%$Ezjtk-Dj$EUJhe6tAUEtY|z_e2G>(YLXq(TU{e=^i=icCEu02&yA46}!AiLQ
zg~5%g`QR{bJM@!12z4zMQ2JdLPN*z^lTMo<Xqh9lYV%!rfdkac+5*p$)#0QN41Wy(
z?3=biiE}8l>>A3xy9L9`Yx>ZoHWNBy`$JG*u4q_zC%fsW0B7>jMDL#33yWN@up!B#
zp<~tw(ZucjV4v0*7BsxBqU(VLi*Q&d#Gfq|2N~3gYk%$+Pxvk+$QCUZOqc4AR#1Q1
zVfRkB+ap6^)x&8|vnQ=~2&cimo}`pfOW||9XwK!$B;p*{g?W3)baD!{w&#(Md4!^(
z;wi2!lKza(q(8TlXyN8`QXNxHdYsMC@$xDKZaqQgB=6AB^ta@=?<D=PK1<r4uai=8
z1Ld#3Nq27FqCZ+M$<iI+ZR|EMzPty7<fULQI|-6+`GHh{D<mJA2bpvD-e-mj_#K`P
zGKTJ;H^B!SuExQUHBm6-L=b#%4S>a!TVZHn7^MD~5AippVEbhqsN6IOQtr$G{m%WM
zGK{lbWW(XD+!U~J(Sat>JV;!PaB621eBe0;xADdxTRIG$9hU&=S_yB1XF+9Hf9O~~
zOY~Yi8`PD)iUym{60Ww$lH9;LvBpRTX7HS!T{aa8eztc-5iuLZ5$ujwdjg6747L<z
za*n3&k4rST$dejJSklnI$#mwK6RF=Rp^Y+*1dpO=*zgThKJO%{_y>}uQYoFk6F_G>
zbI5JgS(^6mIH?UfM^_KVlE!~0X_ZSBDe4qZO+o^#epNup3u@_&OD0LmG?L@8^VCxE
ziaUBL$ZepQKF8jrr@w+>!m=gsQQIFr`D}sh!)Ae_<z7(ZUM8odzThYy1bV|ZfV)jF
z$md4Fh-*RMWSb9)19m`NaX6g)5(;-F<iXIznec2>28eS#VT=Ar(Bi$=wa>Re>zgz1
zanoX`<8%3+Y0;o}VIFjb7QxB?IG;Sp8P;B&0)au9&@=roq^~#(eIJrR|He*G;B2zI
zfJGpoF%jlPJHV!6hVZyX%-$JhGSx;!L0i2-tew5A!v3!<8`{s7#Z@b_L^Dfrkd_wx
z>JbZrT*ZPWKZn0piz2@hl63X^6x#KrKV5jclSaC!Q|*ZZH1c01nSQxQy+yM~dYla{
znZ-RY?%7lxm`X+;GwJ5fv*i2xJ`KB*P6~SiDbgUD5-F1m)}N#Bj<dA-jF{}#+@;Aa
z?Q}cr0(GCSq&u%4(YCZmfT=6N=~x`d{*SXfa3}a<66_up0>ka1z)DyIvCbP|^XCJw
zU9TAEWC?V9KLxHs)`N4}6)0(pf|KXA!m&$npdIf36U;9{aG?jROgIFq7l*?s<xuG7
zvkv}@7r^dR37G#Y17D-laPD_5<a>BPKmK`IZaW61o_WyoA_+{-pM{o<>sg52aVA~v
z0Y2VqKv3QWzDuV<Qlk@CZ&0KW^9MoO79A)_xN2TxPOLqrLiG8~T-IV>DKxc?6=gI`
zBQJ?qQJ+<xkhbKM&`?Ukqn8J%rnOmUo0Urvvzi4MC87!+3wi^^WHC;cBF@hwMV|;7
zH9COLLB!-SJB-d*&n0Gch3t>-CCeLPDvPS2!ia2=eUML0wUN|!`3Px6RM3%q$LQsy
zGHM9FK&`b^G~mCzAT`Jro<B~4vmt4qG9wJSP9;KFb{>o@I0Nx7kAuqa!(bv&2KEz<
zK%(A7SUEcfrfo<CasMnxJ`xSGnWf;YD8QlY9B>RI5K#&I6_<hAt7l;U`x=z$r-A2%
zYB*fe47T!npk-?%)Lp*|)#GnM-NqEyAAA!o=^lpmx8ZPFvl4zL^Etg^1jxvQg3mR3
z$nH*r2*o3iA)IE<S4G1N>wTa#&=o8lUYM&`Phl>8gF$=JuZsC|N`zXA+3e2B3Dh!W
zoKWqYA(Rw97XGNL6(0ZCLlY5&Kh>rb=-MlIy{i^VzO+$#b2@48`){(`2<raM-J}7t
zC@$?5#qj&*kNO96rtvVjmM$jy#2qB7??w3c7M*#YNS#k=sn;%@LN>IMO4cLtx}Qv5
z6NsW!E|6C71^6+Y^8qx1IU6G#Jd4i3wl4|rUiLB=)nA9wi7`;LC=*&vRe)>PLva6k
z9fqy0gAVO-_~?)W^j{p@u*rv;znkFD*t6h$w-!9=x?%5wSZHnV1Z%U$(9n-_iTXYP
z>2|Spcb<Z{_Z4zClyfidNf^62jC+;qAnxuXIH*<umb>|xW6*He7xoa&2c(1jvJ6P-
z(1sRP3}^JWLyy5msLaxW8)@Ge+75<~)$iDF_{zF<B8B`-dN8T<DjVq%!vfyT6W)IR
zPY8-VXFht}E+IzauvlZ15`AyA5T~ZO&=Q|`+M}{mIF?yR4^10{Pa)A%Y-d6fcTcA-
zwG~vfVjRgR?xi%RAX;*{k``zA(lOISniuwxzNwZ|tobDxxry_67v3c8ohRr-LjsLD
z{ge#dzR;Qtm5`Ef1U~R<&oJZ=T-194dz6boZ^>1F(0q7(w-g#O;^2M4Wl+yP1_4J;
z12i;%#t1R|H<Mt^xN0bF?1bQf=`gpu0hFi}_Q@B*d-HneTKF42d@F_Z(N%EAs~GmY
ztA+H{kHPo-Pw2<<y9Ei);h91wm?idr+LSh!=;a18FcC5?UWEfuq40O|6>xP}2-DN^
zS;y!9Sh&#y_+DHDNlQ*Z7xx2JEv;Z*HmAXz$XsF1EO}u!JI)SO&4YsLdcrK*yDa|b
zGjo#lqpzc9vvRYkq(0)c@YH89c?2J$LmHQb_zMxVFJ-jgrKw1N*6k50kIbcltSDOk
zdpT{^38$vOB_y|Q1oi3Kk<zHC6fE6Ae!077;m=&^$s}@F8cANOD(I<aHM#m;;%C&`
zR5q-d4xQxp&KK7qye=Ll)s?}NQB4p554i8J6eiAp1k>eP;7j;Dh>@@59jso6y8H|r
z|6GHXF5aK#*X_gi&q00uXBb)D0hi2r;bwXlSbIK)^?9$LnzK#9G=4&M$zTleiiO!X
z<#7kQ4VJCzK|w(lSCnx6_l8_JUp*Mj(i&lPXdB2qeFsqY4&=)aCgfFv%7-LKyxj~g
zdz;~!QYYlb$HM%p44QpnVC?pEsL}Nk*WB=j6uu8p7@Nej6hmPVg+Pz87if9+FwgCp
z;Cbs96Hglm4S6HzQER>Ei+`YCxF(rGe`wO@iL<D5+%&<cAV%nwIwWYW%OT(Orzod$
zDV^!%=daa754OCZGc#tA$(Vzr!q38ccW<PEDa9n)afrUZ4B^bvSn`w)ri6j#Ny@dJ
z9KM_fZ<RFeZ^{SJd%nL2zXD~>g)law6Y}rggU=UF!KrZ)xM;|8P*tdf5T#1U+m#MU
zZ{=|FvPWPY`x4x@U4VsoLs9#J1g3uPg_M48pzqCV2=bS~@8jOXGo4<rxwnn)M5OuN
zF9kN~G(gB{d7SX<1*{zU9S;6YfW&4sERGwDOI{v^rXe!uI^-6!clZTuABJMw%mTR1
zl;PfjPB1Mphb!?pa4#box~lfVf`B9N?BaE}r5Fab^DaTxz{wRV)`miDi>&Bg=4RT`
zY6wxbHA20eCInR|T)MOUiD0+5hBc>56?rO56b%)^D8Ok1d5(A{miSUGgwPh@wq-Vb
z^z;*UxMc~Bjso2~97(_HPf$;6F@0S%ns#LJXMf>wI^T4aGF`8bnNd91-b<y9O^xKg
z`wE@9cod|PUc-os0$lv`3a;_lZ=LWL(u1V1y7vVfNU4GCgE)sU={xi{xdl2pQureH
zD*P_`41-2XVVY$Rkmf5e7IRj{zPI2$M;bp!_QM?A(dhlMKgw_Y2??oxA#24@tlgx`
z{bzq*{5>svXE+KkJ#7Q;A{9)(EQ^6n!*Jov{@B_x0{a<h;J=I!Sd{7l$+61VwskiA
zeMfBZ(H*QMRR%v*W<zJma^3;yfYK!lD*t;0V{#Q>ltPa1xGEfuHJ*XR)|v1vd^$M!
zmWei38?pIgQP46tE_gfVlC{)wGBNfeiHgr<)jmqJ#W;@6NrckTNx`(wB8fK2aqsR9
z86o7SH(4+DqRDs4sT{V`ur_x}YxE<^jiW`aJ19mXjN~dZX<%w98LT-+X0`n<pR~JA
zVaKW=>ADy`%RB<b;l1F-J*{`$4#JA;Lim022OPWh8{Ye*ftuV$P)Ue`PG=b$kf?-(
z{y#yA-w(<=PQ&60mtlKWIrx+hz{9IXqEAl?C~_}NJKs}{e4~t+yOc1{Z4|!iuZooO
z3D#{_!ePL#;T3OSnU4;B4pTy5Q3oV_>w=r1k3oK9e@xwa8h&=&gk@tjv2&;<UYuIZ
z`8|plW%CY7j^6_HZ?AwK;~+iviYT(O8jeoq?{-N~V9xv_prw2QjNEH0#F5;6o#+NX
zLU_(rw1NGR*a|&GMqp=|C~gk$r^1Y*!sdG+lsD>?c)e<(AlE*Mxkf|?KlBR)iGP~H
zpWZ0iB<)QO*Y)X3(LAbT2T8eVBTdR4Me6?w$>Cl)eGbSW&z~i<VNVJ<-0Y!jojRJe
z_XW&}{sEpkEg<zq9%~F`(OXLa=U-{(?nMa<3b+Arx)<SPU=MdQ4#VDws<?5a77mOX
zh`qIsV3Lb8Ha|TJGMkNfR@@ZzBp<+s2iyyFY!uH0siE!>Q{2Y+4O=#}fI+@K)|F4e
z*_8$;e^>!Ge*_#duRrd#wm`cTJ<xN>7<a|U<D<>fam3!oaL9QqF03Diat+!T^mY%N
z-aP>id8Y#9=%a<9E09eFyq+wJ7c|O2o^xYVO8M_mQxBOLwV=Lz9CRPbf>)J!Ok;Tn
zObt#HcKhxZdY7$)8{7CfFHF)rA})i1?hc1dZn1*iRSo8<Do-VgON7D!D`?BPN_uf-
z2t5w6p<_xR!urK=Bq~iLA<>m8d(0?hZwUQbwvl392sFZk`;mnbl1TbanvJ2<H&qdz
z&VB_m`A+lUt#8nhQx7iOeRZW;0at5qPEk!6T;u)Dw|CkhY`hG<@B0k%lLz3Q%cJqw
zis!KQq7J@xd=JI9)v#pra~P)617S+4_@sxsP%jU``v+w4?n5~|)p!#^f9j!rSTW4t
zSu4qqvFL3ni6A`+6F3*3$Z#qa$|k|lrb$?H@Fp1Y4)i3`cxd~60rUr|;S|}?*x@!9
z%732~*=PL$+y6q~`d$O<)ffXB(k<+J++F_MlEsU=2V>D289b9)0-qQ8Fh|QQ)^_g{
z9Ctg+e8$W$`_QxqOwI+d4O<0Kr<pQz9T`tP7b8UqZpTGOrnRxYk{DL%xt_TcDX>Db
zWwc_~dh$ABFNE*OCATFh!sC6DC@j~HTJGjiru=wXy*-T9X>BH#q;_g*EGLheR_d1T
zhntLLF}J@QvVcx-;eI&BUt%bgs)e>44X|Nl8Avs0;2jk;EbPy(ql*J^Ny8^FxHpBn
zAAdp3hDqplr60PiQ^dP&HaL@WYP|ao!IwYe@nY#M7&p8bem`D>MFz`|&6<k+GX4Xb
zVdJo0pdNnpx5k1(TTJU)h>q_@@Lu3F{8{~w^IisG)_)Um%{6-zoif3(xApMv05cq|
zxd)aS8epy$@A8%iu*p9J{uLYs{ny)ITkc5IbzR2>o8AK5Q&Lz_mkQTk92S0kIxWP0
zaDaX-FW_*a1V}i<RU}Vc4z06piEC|)sq&bv+1S4zXegM0nQ{}WG%GM~F}}~5?!6I$
zq=HCc#6NNKoHdlQ>!;wbV+gtU?xS|2TpBaRojzutrZlVdwC_L=rEniter6V3Pb#36
zy<fqr_Z7%5{seW&D!8X(2x_nR2QNpw;4JX};F{eatkoZao%-tBxlseNgLUwtzZy<1
zQpDoKL74OMH{>sr!X@J-W7ZxieEC-&S5H*of6Opm$q8>&>*0<7Mf~fw5{GWE#9ODj
zLE&)&%;#B_{v$2WaJwN+<<BUYqE+}ig1d@}hvEADCUE{xJFJr_2K(*2%bvavleZT^
z&VNhLKKl&(RGNvOvlC%Wa3lQG(m-n+U7Rz)5^8I?V|k7q)OBm4NA-Hp%)bk3Jh+4U
zWe&_L5eZMPa)<k9BlvM)Aw;_;gWV-Ps(<)IwD6(|75KjqPA*u$v{!bSr_&@MWp<t*
zVf$8O?QJFqZ``P`R)tJQ{}T2c&Y*7P!`v%Bhbjt2l1_9E_2v1J_)Qf}o7zC1KHsOF
z8@!v~ql~L~|9H*cFR;kJ9u7}8<M+7^kkQb_()+{l^RhlL+A|6}lXdVygB0#cu*BZH
z$>>#}fRm;hp_}szTr^u1E6o~U_?c_)IC3GnJzsz?FONe*zR#9evK+Up9)Owky4Z31
z47eZQtema3Xnar(3;2$^)^Igejc$X@Go8`uYXr<x>BpUNfG3|H20T5F^VAohw*Lye
zr*xjp{H}+AUpNcuge$iCUWN_2B0L*yiS;Gz5Ua<3hIg^%D-!i_^JZC;Co5QWE(1oL
zXog>X-{FmK97|a~Nn|Dj(jBYg%=gPZmb6HU5_@*ivmhjoq#$yY7*8^ya%S*ts9^cD
zt3tC=k(x$apywK<WN>7sV9^&%1wW=z#PbwN_F710>ZRnV=t7}VPibJG<Ymo)r4(qd
zkKU69V3*xklrWIPHw(YO&LB0k{VIVRsl<6as<=MS1Ow8C;QDSGR9id&Z{AVG?g70Z
zQ85VJrp?7b>8Us(Tptr`tT4)aG`{00p}DHI_^E6v8t!yJ-_2?mKX(N#@|D3M|BNwL
zP8*+ITaC{4zPN|Kr#hP;@2~!bEfs2L%zMOUsd}i`xf(48@_dihaJ)9^A0+&h!ncpN
z;zAD*zKR`*E-T05I139jIXM#t&J)-|twfN#&??$dCWA_Kia21_Yi1NY2;`Nnu@iMt
zIQM!JWY77@bkBYT@XG?PrJCS!lyi9YirBNMYw6DVQM9AVoI1x3A}_ar<aO^H<<GVd
zA_FxjH08LUFlH4Un#Obc^L(f>z(+9YIwy2P3fWrxp|@q*C{#X)_LYk{^Ot)&bcbPp
z<rsW3UkkMj#$duvbxceciB+viSTR-_2ltubeSe+>_xKD4OL}48*wJ{H`!}08pImYh
z!|IvV=+UajIW+uzX!#oce3^oju>up_7UI*9PH0_ijAfTyFj&e78(vQ3xpp<wdTowU
z+fA^mU@qPao5<Y=^7wJ8F0MH^9u3cMhlZv++Qu)#TNcKsw{j$MIR*qK3_z39EAX$i
zF|O$~#IDuKcra)(4*jZ!_xv)L(rRPeEawBNM(5ydv;=lE`oW^McOb*Nkh{amK;4=_
zkDm{WxiX&Ac@j_l>KDjAcn&Q6B`e}@wwT*1>=HH|m=3X%mqMLuD0NpI609WEX}@G5
zO;kHTJ{irzosb-I35%duvuau$lueT--KRmxaU?mvhQd`M>04zc=S&@>v15(Uw^I?D
zb5$|rq&x=8tKe=AeGFG^fOlOJaDT~Ie!o}5{r8m7`?wR{`Q?W8s-;l>?IE~@8~(p@
z$}4pv`mHg=)o9Q2OLjQ(!6e+aKnE}Lymareftcy;j*e$NPy+(c>{5SR@G}@6r2?j=
zx?!ln61;L}6BgX!jPEEn-0n31SMm&Sd`%Ee%=N;dxl8f;aa&xKWq^}8`&N##TbApr
z!t>uIV3E^a%yZj_CBKw$%_vWJxu0Q1zA2VJiUBq4CouBAi6E4W!__g}@Wi@+Eqwn4
ziuds|#_0plKRSnLq@96Xe=ZBFHr*4}FPRB))}cc4EZ%>ft|JP0cZkkDRpz;iDC!%U
z#=B2^bRc<}o=-QUQLh52$!;}yObw@+emP_z^OnxJ#?roN!E{g}nI0(1qgYD^l?v=J
z<=qU7^d5o2TthtXIT~%-c>l3}BIcToMAcp+&dQ&GHw!$`Y2*Ub>g25QP7$_Vbj0#1
z9b62v@!)+s?i<*`9nOpJ%pY?M;tYCmjyoFdw?O&mAWT_cfN`}97p~R7k`p>OIU*dt
z*DpeiinW;TE01(zB>pV-!Ok8PH2NTo|M}XY$KP3)mN5V~Z(D)g?{9)!?ME0}?~E?L
z)zPc|AcT)|MlEKJ877GkVpt7_szzeP>onFoFAHYIY=uklvN*{;6{fGf200T7Sm2W?
zhDE30%z$A~)NvBRO}B&j_A!uGxQ*)0X0W0@Sy(f8J$xr?N-6s%R(CloSSpSXPmu4&
z=hLMm+dGAhKA1?7=hX@BMo_Dt8|{?)Lte2L=zAAuO8Sc^)nzG0zR*Jb;_>J_XA;h;
zTaKpv?9k7C60XnZJ{N0-WiFN&Npta9#spltPZ<NbhcKvP4tl3~V&7G3%)T%VYu3sm
zo92y`4=3ZMUD}w+`948zmbfu*G)5`}<K)<O*tX6cKX2}ZgI2CMRWAtZn;2d=Yk_8K
ztg)JR8O|rJ!PNb(sH43PbNTG+?!UQsd;)iG=h)+=Wt-6TK@|GU-i#({ro3;t3U{RM
z12k-aUKdx~x^o2TSgnLllRVLTUMu{JorZS17K*}G%*OX`B(ZS%J@_BL)7gGBR(~l1
zr-emOlb0?u);@%s-yYPpD-P1eAB7Lwr;>pegBR;vS*zCBii%6;sc(P00JBvnnKQdu
zH9SQU(N5(ZAFFATMYS+<k`;;0r_<racv^q5n(P)9(aA7LOs|@b-*!#M<DQ%aF>g3#
zyx?l!YR)aNH^K+5F6hbM11g?*ql@cKR0~x`rG}x{j7n(MJr83l*Ws!3nb^MH3J>o0
z#a|!faNAYRrs&qiemm7Lgl8xDJ0TtvH-ojtF06f^it`K(pzt6b=VfZ4?7ML8J*$Gu
zYuj-2i8yrc=IkDm$=FuU{T74xyTy%N*uq^s57zMe#q(GU<^DS9>yjva;R7V=w!$;x
z<9xsMOJvUZBJ+nR<NJ`S(9hTqE6bL`u#(B>dbOYM{ze+u4O#`sPPagBfj7viToYCV
zoM4}(W(xOuwo~%4B&Zx7Nvfs&g_Pw#SXoVpP`Yh2)p&grr*0hypAQcM^_r7(W|9H<
zZjT_bZwpze>66d87UAQ^^Q1Y~jU2B8Q^ejnni>6x(&9Am$4d)*5vq;J*O#G^#Vq{S
z-whRcZZ9^(4Nsr4;qF&gTqt(L)DUajo*syXgVy2qWG!@kx(=(aEyar-Ytin+1e_&~
z!r=L<F{3~eYo)oDU_am%vrTCEbt|rt2*(X${IGy$fv2@Fd|#G=)6QAp`%dncJ-Hv>
z>}iCC<$&+5t8)iHF#cxzOe!CY7n=KFad{|e++2u$AFR>ol^ur2D`36cZOGhq5XI3>
z_|A0>w)by`EB$t(^0iSodYvJfrb^@4C`Y^?XNZ3fyo3u+jCfb$9=!H4!U9Dx)Y>h9
z_ZiE<s9PGwIQ5FWLd(ELw?8>nof5}Nd6WJ8IpC!{1T?M}vd5o`h1!~<bjPGbxHzaE
zwHcokYV3zxHtf1j61KTia-dom+h5{x(}FqFCz(JMbLV2=RvWD7Hyve1`=RZ9M>Mxx
zi(?nNV?u`^&e^g7i|09Fj=ltEku1Z|nQJiiga_U|KM_-QufVv@2y|cOg{xw=;_4{@
zIAV+;=1FcwpNAvSjrXOdU0IEh^R;ovPIZ(xuoU|`cVQp@K2AQIh}CO&4=Z^pmR=9V
zCA_N^{NH(4uN{cyQo&die-$n!hofB<_ksNIL8bT&*g3`<f2k}&>1)w=o^yeI);pmj
z?*u+mybpWk4#6t^?)&Y78qR$X1Z|?5tf(*x!cU%r?h~fyVEc!;|JV*YF&i}E+l3n+
zykJ~t2b^x%1B>pSgR-DAaP0OEw&$n{D0qKoBPJ)30q4)$8PUd4vQ9JE)YGh<&-=S0
zehE?MbLqgFHflJPNNd%0lkeoiBwF!;+J+Cj{L8+Q9&D9I(^W3mt*gg5o5L}D%rYE3
zR};hAOwpIWr`pIYL9wwtmhlcn{9Dd1;kn9URTtbfzz~NV<lI7|FjU-k0KcsrgWnYf
zp<KrbyxgLJ0k*3#ZjKw~>2R*X{XndKxEn9vbX1e_LxYbCa7C~>mf23k&RB2mLSBjw
zj5*^s+#35{X5w3(|9?I-1Hb=z1a4%6wb!}ZWKAR{>P29o!D75*9gABz7aYq?QAMT#
z8tOx^=KBG3)m(#zH_kwlNt$3Y)EGAo$%BnnouIy!&$1qI&RKsu(7SX*I2kXA;X?+I
zwp%PXPbh~-(J>)YAr{VzsbJ~9-U<&q7+jxOBxqk+#n$H4FppXXaW@U3o5K&2r{;0e
zWRldHB#0so`%u5SOXQ*>arsuCCq)f;Kx^B^(XVl{a9WNoI?R<rbG0$3wc8vQJX?no
z_`R%U&rA%@-iZIl(V559^u293q=8ZrlB7gZD5=!h&z&TsLXsp&ND`VPNphM;Nh+z3
zkR+L@vlo(iPLj+ab9|F2gx`AK|9$*%I%n^-p69-=>$d$!{LQgOkmf+7`NiWL@#kjy
zP*i?dhY$}d?AWv#>*maXV7m<y4=qL2k7#(6ha<rt12%uNaQ4`26s}o{{0STJGtdAr
z{CB3ZBpy+HXJOENJ4{$G59`OirKj_!qkeb^{@i*>3x0HiTjWn_RN|diy%fCpT8Irw
zuE394s?*s)`9qJ>%lt%WeSJrli#d}#VhZk_i$+evZ$aB)1%lU9(VqGB!jADe(0|mA
zR43R#>$IxmU!#=7M~aa9_epYf;3isZ_fzQVH;)PirBcQ4!D8{i?t)YQJf0)=6|ayT
z%hUTUo?P3D*`{0&W?4kAGtyM1#AnbAL(|ymCIfb3V}z)lyhLP$O{~Z2*KGQ`DrS&5
z1J;|D!0y{f=vUaIPq_iSZ*Ujh+c5lnl!g$EaIE>{j^vlgxUeG?g_&JpbI=u&N5sK2
zaXh+z`$B`4u7gEwU%ZMLg9p;C_@bPJ(x}Dw{bf07?#+Xp_F}x(nTp2BujKY;0~D%U
zp?oA4tMA%ES;-A+=UC%d$YOYh^10P*Zv=TS#JIE{^skFAcKln1t_Iv!)36bXk8l@q
zyb+o`HPQX{2z2mg0C{$Yb<fWfy4()S`TI5PgAJ?>{-oQ^pQ&+%4l-SM@3_bsmccrh
z)|4Pv+p0;zevRZzuz0eWKaMI+_d@pH9*`?JRn3DR;<!1s?C;X8LZ_ywX!hJjX6M==
z+#1y)#?1S}E^nO0jF2j}P2C`~h#4Tt{mrbpweL35H9RlYFFV1$-JZqrqx!Pw+sj~?
zI0wFq#z6eA2q#0$QPXuDijT~NO|dEtDs9H(Vrz7r<%b4~RoJ)IA8S@GgnREK><{D|
z3N{IrL-=ezmpe+nTS4yGFnq}3`Q|cHr0}qbV7?O?k9n8P#1?N#GBBt*3UN>T5%V=4
zUAa^3Hb0Y2_m9AH-Tlbzu*bB`wPf;eGwv3Lpzos`=(;E2`91DWdw+rC_4Q%Wbc!}v
z{-B7_>!H2=9VuR!3fYF?Xz)*>=zu?Te9#CuujIY>B<{E`i6aM91t=9Rq5N-}oRRsR
zdR$1L2-ON2W2=jjdIeVTDOq;>-3{@@abMPKV@Y3|eTC=Oa+pi^e767YY-+QTm!xRk
z6nBlDNy?ntakQ&0+wpm?=zC0FEZ-wz>dShZetJw`nOp1GZ#DhXl(CvkAMOm#>1$vS
z&Syf;lkt4rQcR_>pjZzKQs!*9F=05p!3{As5tuY^3$*LGo3)htq)r*(;f9^O`<sKW
zomtppZGa*lHxxx|M)x*PgdgW@sje&Vr%x$dw~jz_S|}P9zNEm3IcN;;jwPG;Op5at
z<F5O_v3Lt+a9LI#pK^+Gn1a+aIUJP#Lrt?aF~8?Jq{*U?=pTZsdk@f!t(?m`cM+;X
zvoUy^fXRKMFzp}z?e`Ec`^P4l(0vSw>t>;5!hM=&N(k#bMW23+z#->K`Zaro__;At
zR8bs4UUO6#8-9Z_>+OU;E7nnW#|Sc0mq{Afc+Qk@6@AqP%J#h6&929cWdAjUQ}Ht!
znU|u3jqDx8e0+B@r5!I=hUFV(mk=T<6z4JZIhNuDzcALUpUtAztV3#V1Gw5pA$;>(
zOf9s4#rf6fb=3i1Ml6NqzF-9PbAdsDC%S|=U~)kk7T5V=c9s*8b{OONY5`{xreIUv
zD#V|4!wuJ!kPi7l4?9+(%XIEad0)l*VfhHVwihLR6A<y_B@N(Cv_T_Rz}3VVJ53jH
zr*<5ANwQF(yb%xX^@6D3jRhI~q5nDoff1vyb@g0~S-TNye%nIHAqfhPSK?N{WDGu&
zfDLhDklbT2<mb2}aO(#O?DU5$o2cAmGo8IQ8p#XRlfj#E8dbQB4E<8c$J2+Y_`DNz
zBcGFw!v~p3*nV=&;Eu&-zv+1Y5>kxcN*AooQeVw%7Ls;CrkA>tl;;OBnR%sXHf@UJ
z^jlN*sr&?cd-4a9v#YRN_Q#v;&^XC#CWo;|?a4ehv67XR1fqraO&{0A;MCd~SoJ;{
z_FK#_p?e}K6Q?2Z>og48F&-!V@-V2<1cgudzjK}qje_m?Qjv={ANHYqSqyiV=3=@h
zca$_rQIPqU280a4jJb+f8sdSzMWZodc?nu}J7D)iWo+(|4fFep@t`gY1E%gka!+N*
zC50pIM-4@e*5IA17^L<J#ShgIY<$rR-}Fae%&!%QU%LfQw`E|_bbe;rZw*i0k9cg;
z1ulR2yRAzA@0;^yb(%2(m&s`MsdwaZVGOD`KO*BsJ&ie0FErGev!@zeFi>ti#czzE
zPkq{n`3DJglKB+jv5iIsqzm#|ynixp8XYhAE{0uJBF)LG#T%bj^S*UB)3~WlZ%^$J
z>(;7?9bt{^!(}~Yw0*W9heD=0-kCKQ-eCjfRZd6v8RM2hFl6ECc>84{E=`?-;d3Wq
zqRB)w|DDObhq*Z1vI4R<9+)2E4{3NG{_M@g$G)6Xs>E48{ntbKED2iuLh(T(9o2GW
zFw^to{k44f)Zd{a55jSg^Q-3R<-t|54c1q=Z(`d{bRFdeSN-9r<<HEI={9&1uog$#
z;_#`*3Jmw)Zp7Pmi29U`z2+r+9>veA8?3qOumCx;7vpS?NqF?)676fyfQkBSJm4<D
zrQzQxr}hcGe8k=1s>^Y@#SF?@7b9)IAx`TJ;@;^FA*SbU%IV*gie(ve`(Y8~SocQx
zD@JlN4aH4PN>Jy#wVF4oHZ4cj3mXfR$h+?_q3)oRv<B6%#JP$z<DV}JQ64AC-S@LG
zR*YkI{a@Q?E6bgp_|}i5+}$mX&ah;M9>||Qal#*~m8IC#Hxb|EGSTIOBknFsfa!`E
z+|THZq(EmJ4@*V*x<dRl=ly=GY-}nk!u6wRunKm@HroU~|1ZSn%!SbBEVa|$C&5;C
z17;g?m)@UrSn~VlA^i|^J5_+|FTAnQ-w~x(bddXV5iATs@aXJVTn|%0_RDf)xf;Wr
zzf(5m?nUh9Tts$o$F9+K`t@Ta$RrzXU9!>eX#i&0f2RfZBVe{bm2$5R#kw{XGQ9SS
z(wsDLQ9p<3Sq5#t(*<2lJL1AZV^~XH(TfCCl&xJQYhP={{4V#W`h&^B+L{RJf2}|3
zTmK(*-kU7Eh`C2k51bUwc%J3%h@q19XR#F0^D@(cfoSKF%ntKi>V<B*#Aq8cI(d4d
z=-K}|dq3<pE8KaMrR<7j(wa`Osdy*Tl-gqR|9Q+*m*GJ?zo)NWiih0CWOpST2Yd55
zh0aR+)*pvkr-RY)Y8l1|Q&DdpkHFL<G<xeGc+WPR<mbmxgHrKXX(Sw-!?~OL8%;<n
z!;gp}Snn8!3yJEe+0z?ST@o<)jyjrZ6wz?R0czX!V%6w8exLAzt^XKIe%u>-XGS2}
zGZ|WZ@APrWW}funyRcu)w6|slz4;b{wVDfHKIJdpeHfx~DCahhk-%ZVG;EIfKsnrZ
z*k-p;%$!$C16#VI{Hg+JZ}Y+8wJ(Lgf7Q@twN2)wVvjj*X3_S26%_7o#4;>$ne^Ex
zlByga+v%<}?p7}O%#g7TcV*K5Hk>`3w2dDA-9}4pX0VBViFA662Q%+y!H%mX3Keng
zqO{UN^fk$l_%2uAtaJ@lZ@h)An?U$wI|gfKt;DWY?#s-wLqe<z40FR^ba(^2gVQl=
zSQsv455NhpMbIou=HAO57-Z_ny`|HzLct3IE_6r3tG;;c9}PL(6?h!!#l18VTx#Ef
zT%BMP@!fUz^#XdUW#Vjw2TUe$|G4pE@|t`U1AVunr{Mv#8Epqw)MH+lD-QnZjblmO
zF;k@<K5@RM?eBq@zi~U=a-WG<&RYvQI0NS{@&0RpJ)ZC#P4UPiR52$6@}1T?pXC_W
zWQxi2J1Aq;5{d~6<6d)1l=E}pM+F0P_j*meQ&*C;Uc0zuw3L$Uu8DdJiX_qY8<}3y
zBs$%3ksfcK!~X5(=lPOjg0Ed(weeV<g?wd5bwgYwS$8@lM!uDzarAL<{$2yt8G2F3
zd47o1&(LAgk{hzOPfm&|zwNnahkHnuXX3N+7=Hc?z~fd!1Y{3~epv>lwo8%TJp)%W
z%klMM3?>yyxo0yJ@kcW8V<mT;Em#A!xLCOObKV^Hmew6AM9Z>n7&)GE5;_3cdpC^I
zbwG8?ZmgM?1H0QsDEW2-y>1oYUA8G)<i_JT&k}SrEQF^L=LJ>`#H|T@#&ui)HyX^a
z^L;ADat5kz^E}vk`@m+eE0T84<-K@$DzUbN^|&<hF-t@0$@M7L7*74JY@ip#)6hrh
zy)bW4H#~gwl;Svl;rId{3bcDquRngL4R-%1AxA^_wS71~IUN-D^^ahmQwAX4V<ok;
z=nJ=xrP20z;Z$ZV(%DT~tZe=taYZTjMrQS)9erEamDO{^_W!mryN$PORG%cVqzkXu
zSNAhwzRE3@(lLvj{N#w&)gjn0!5v@kZAH#5Kb+)e6zRw;B<d!jEC1|b_w2&YEwO0q
zoD22TolxNZR?SE$j&1UQvlH)|M=yr+0X|PUu@l<q|47X^3EfQQ<AoXLDK;L4>6bvn
zo*xd6eZJV&^)n4fFUR7LA9O>0Ctio@<IcCms2#Z$oskXn(r*+_+*^j1JvJbYdp2r0
zuWJ3{4JfnQMv5o?l4tNq5{IT@X`3MmF0Y_sh4tjJa|ABfnIou5A4w63B(BV$nDxEr
zzr82P!-5c{x|5{cHp-4IZl>&SUxluNbLhjcXxdn-MIFoDOInw26jydx$G*ACv(K~g
zss7ee(YDNsUK|f*8jCNmK}u56NNyMZE`1|T4?V<$I!Cr?ss$_9b4UE_wU15Mt9Ckp
zd(P9|RI-}ZaZLNp2E5ow_@`!$$%~ew&%+dash@!ppO@p|id1YV^~MFiDLCC`2Gd__
zVZCJpp9ea?mhUK2m9=qwYBIW=Ps6?(10>naz$1@fsDDjHHlOu1tun&eMV#ZkSs%ZL
z%)zAsLnwDig15yH6wir={N>9O=DC-9e3Q{3pNnY85!BZ?VGid}&lu7V$!({ocCscY
zVlTCX@ytq(*L1&7f}QQw_<r&g-JNuZ44(66xKSgS{wI$|)``gf_fYaG-Hv*nUWug6
z(U`SDg4S*;=uoHv)Nk42v#~1F7pO>z_Hz#i-xE1rZ<WOLI>6>WvSvofJ7rCYY9vjK
z6f(l%#EhBU#Bk?y@zbzZ;{HYDOwm(`U0GJcBZ5cSk|_(=l>0qdz~XPBT<>G-;rXpB
zz^R%0WXDRr?OckeyPkNmARcyo7hy~6VhqwvL6GeN`0ZYShDzSIT_nNMrd>G6Ju+9X
zdSbKj1gOXDL*~z9Xk1U?JK=HIFv=N!dW}a9ejfc6$(a+biRhSL1QS!vhOF2LuV4$@
z{$>WnQETDiIuo5o4x-<#h1gv`5(}0D@ON1h+-6wd>Cr@}op8m;pQ%ur5Q`H@p*Z~@
z7vI*opqpkAPR5zzQ_Lvt-qe8PU@1*)`bA@;3nA@efYa0EaP`6ysvpVc<tO?Jx1Kpb
zm$NRt+sA{Q*eNV9o{6gV6cYKIzdiOTy^d_=Ose6yep7)|mI`9onIQ4`wkgcqum{qj
z2eV}tcTvr(Y-*oXFZ>r(&a~n#$=nwo5X#j1axTyfS--d0Y%<RfDUF-L*<GF0a$A;G
zXIMStE1(s4w|h8xSdGSB|46h~%!j|3Gd8AXq9|!RCN*UtmiuEBKDwfeCvLsJOL6s*
z5lk25U_*aC8_8dV?pBdlT4jhkoH=~BeH6~m^MZZn7!;&Tf{&&OZstycCx2G#={X!d
zSu_eZ@%Nr22VV-zVZE+9f=6ld9hfHkQu#c8eqTJ+)J1#FY4Ug5MJv7x!0DOX3AL;b
zepwAc^&?w6V7WBpg#i={_6thxzv<Vx-(;hxioeA{aH^{i-f-4WmQ9tIaj%<T*11ls
z33?}SS)_<CrOh-}M;DJrrc>H0sj%njO1cu!3zq8!qIQuHyTvm%L$~}CmLJ)`dQGe4
zPI<mp?L*?f$|g$s8p^wb8G?&VDl5_8&y@oOY=wG*=yP-}D;9iZ-#_nR!IDCj5x9wE
zJmF_JKS!9q;jZAWNtiF?JoxFe(evz7w2YgG;B+S(_sv11$uJB}o4^^ROEE-Y2_|IN
zz-Hb&WIQZ_xmh@tbtXamwJof^KcudHaY+4Q4ux(7xM<T#k{{iXIcPZY3r8Za!VvQZ
z$Dz-3A83x-g6|c+@Ki~_wrUNmnq`V%-2bum>0&I+jeu#;80^h&r4L!$>l!i)4&PVP
z(D+t5GUpdPG~qtK5KB5>ngcb1cNE`S4wENtqajT-RO=H5t<<}+TX81X|6wI9d8JQx
z3msVGp&=+cR7f5U7sSuT8npVO9J8LA!#)K1Q@`@pMB7zhr2m2YI(s4^%}4zHuab3k
ze=Y{aXVLU~9b(X{SdkvhVqJOWM=Hq`&FVj}Hv1Hb-VQy{Vbx=?%y}SlcyGuqjb4QG
zDhJLz<$L8pd$GVL1&_{f-|mJ29OL&)izi7~JE$D#oLd~Uek}KyPT*WITLc^q#K#+)
zANj`)7eB9sPxubFoAEtsNHVs)BzVkR3xn7S)VA5fNhbxN!}emtj|Y?)Fc+Ws&XrwA
z!O?zQ(0zj`D(hWP=Qsjwmex3#w;lhMr^BCnCY<aCVNWrCFK(WXfYTDlS#qD2oCo}j
zc=v*T&L>=MQ_$Ik(B?CkW;u13ZCxcX3gkXNb;f%K!0`ZEirlb}*>@(A`>R+%S6+iV
z>Gfzpy$-(S1`C~*`)RUbyr^EXPG)uLr>ObfU8sBLD%n#sS9}@I$mYX)l9!JVrwcjE
ztj`jOUiWYzi}&U7oUe#E6XRIqn;WdLxSdUFFJsz9dhE<_<<sE{lZ3j~Wys`nvu$}P
z_~zq}W%aQb)0vKy(|sZDv<6y(r@`c)JMPcqUc$*yIAJmgamP17?v6V&&YAOfxjmu|
zEJA>m6nfeM95`DyrZXRL#a@__JqaIQPDDJPj~;Ne$3U$OSkZSSW}TS<LuF0uO&Ea2
zO;ZrME)!C5F@iZOY~}k6SZPv;U*L165q#&Z@|<>fne&auczDXY(DhRr(VnY^QNfq!
zro9HAmGwfu+x^f+DF~S|7joIps7MkkD6D=-J~eT4YhZVr==7$L5xZp}^Si?NVlOnT
zvJhmJ$wHW#K6`&FjTOg4ORQ$xr{cBYl7rpWlc#cr5cge+du+}M>z<8f-sB_P`rVJr
zc2|qKd_OoPuaNEc8piz0Yeb_~8$tc_Od;xAD)WvkWe&+w(QH8qO20?I&V3s0%9%lD
zXB_wKX25FpW-LrwhG~0^;r%BA&r*sYhn<+UIT&f1S3}o%B~E=Efqi_=fApt5;%*z_
zpYs-^^ZQ`9Vj0}C#^Td0em@ga(e~aC(d%B&{OzGIer|x4l|QJ|$Pg1IdSWZz|NU#q
z=Z?M)w5LJ^v3)jR*1BX2eBDHeZejTOt%D>%$%u5&f%2zHnpJj@7VZef+mHM+5FJsp
zDwia;=U~_dMsekQ{t<dXh~A^g(!WR0l>-aOb%ueMxNbCw0|&u&`anc;W@y+pTh17-
zp{l%e3JVRO#-4A*hptkVwTfhB4(ejad2`7y(UEEUG>VD;)Y%fnNz7X8ExcHx%(Pu&
zDB`_8r5`RJ`KQ-dMW;RK-q_AmQhKo?HBXt5Cb3e5Ks@1l(8nV^cxH;TWgjkv{=i%u
zI_`l!x7I-ZAHk+61?|IA;LUSBGL1c0+tCkqwob=qCmVbZ@j;W?cr0j>V%TQ`TpjBN
z&(2M#(@cRz6L-)4@r1+dc^F*T6RKNEaPIYRoU~PhYl|}HDh-8b8Uz&&DZB=0AyS>s
z3|E-r*^tL{=0g%5L`6d7d_HxU^o3Kb7e0^j!kJmsq=#!1eXfqCB*$aM1@2rL(+g)W
z&%xuX^0;;N1X)$?qBz%7QNBkD$*r!k)~Q=WY{XrPULGm_?LQLzpBdqAixrD|r%&Up
z<EiGrA=d0$AYR_HkmWz0$(j{Dh;GZ?3I*Eoa7a5Omhyf=!GA}Ye#s_wS5Ji*8(w5~
z85u(J3^Q4e{T(dp=@qef<3ll|XsawgHd?&P@1U!!9nd^}E<Wx~z|gMC;BqPh6_aD}
ztTYx^UMAsuxHlGdlfu)UGa@-B@20r}&x~bZYUL0Nh&P6w6!=p=4_;g?@Md8uc4iL6
zyMJ-`bIt|#=VkFbryl|eMqw4-T@St31smc8XiwL{Ne^FiDf0q)4S~-DHC*+bkAs0y
zoVch3tsfb<+>wXXnZ}5ZNy7c)b97%J3i?HyO`Y9Ion=GO^?e4UoQ-;7Z!?v(DZ+S;
zCJL;66Wq>IM6d^iZHwb~C+-QzQYO2r-B=Ivd?CSkrEq;<PnNfCFzNkj5;}si#kgvL
zxuq+@-NBzFIQ3_aUB1W?To#GyQ7OVVm3wT`e=-WTEu-4A@#M7ZlPLf1ka%T{v*0)Q
zDU;W<V09zaSk9YR7WS}=^)`!R!F=D_W4S(}PDi3jGMjriV=?hWB6M2Q;JCU3$$lx&
z)lbE{v%tuKMsSRuiC&jC;z?`{YQHbQ0oh8p`*420o&glkb5C?7&w~cU!@ZJ!m#0p~
zd+uUWThu@=?xdk$>}a&ybAe*;6Wa2n6cv(KSaM!%b)yz$PMnX>kb&qUCkMlAC3ODR
zK$Q84v}er+vg#5K&x=cO<W>OQUcXG|9`Iz=#RC-Fbq=n+)rR^CcS%ykA<C=E6K?T&
zxyE!~`W1RraBbKkMsLWb&ZfJxGFF93{C6<r1LjoJKc2)b4V0q&pWtk|pN%|HOh+r@
z#V59o?CG&wvDdLl)WbAQ7@L_vVb46Mj&~*lZg&%h4zp*T-DZgf%^Bodb3zQz7$@rN
zX%x>m*|4Lc9*Z>TWS+IEqKykbudJAg_P5zsb%?V`jqD)R+lGs=lX1jlC!)tl@zBW*
zPd=@}Yv+^bGS?IR;g64X*8J~~2U%4*$|~~V@{H%n_LU*?&3Zg}y9|AYt6?#Bp8V{(
z6eq*Rz%w-qA$^jOa;i6<<Lg2}q70+ey|A~?1v57Gfa8ZiOxc(A|BlsJb-<zx(-3mw
z6&YOa&3zz?U=?f3{iX8s^^6Nx)*xYW#8mtj?TD6NGbpe99W^gMPdYi=hd$vdrK+aW
zJs(T7YGzW_CjJ{#ogw);S;lQ9+O)=WJk9PK%0_>+p}Kcn$j7w{Xsn0mSCUUP8Rx`X
z*X~fXO*dTskV}<4y3vXwUxfTi<3#g^#@y2}=al+`U~2oXR9w}N%#LplX4eKOlZxkH
zb~yBi#4hd@XQJ$8sxPb&<`avOqxtwFX7C(o8dg8vjp~IwD|RIxxAjNxGvacd!O6nH
zc7E3>19wCfLb7)!+?1`can3YMjF#Z@oiMEaF%>J5xnE`+ce<{R!3TF=w95EQM3lqw
zv-42;-z|!=Nak}rTa5Osr6&KaaQ*j!29~K|fub?KN57_yXKl2<Vh9e|?ZHZIe#Za$
zjT)myVR-&x%-(sG7HfOr@g_BNoVSF{(xIHK`k4~8K&Uk3{f_V-bmr?!H2P+V!xj{g
zY5Px-tI3s&G8#q3mqsJ?PIq$ur^0t7kz}89lIVGbI6B+_=U2Nk=_w`A;6X7<*z;Ma
zc{EM*TQiAN-G;HoBab=9+?aK}>CE|zU0LgWLmKk0JM;a%SI~0XC)#E76owAzOEvnr
z)g2!7Vobn3mcb!TZ~F4xW1$%ue=mh&SRU@_EXNzy99-PQ{W3@6@qJ4GzOVB`{{0LT
zC1zrO*i^g}LNR>+pTA$=Z2u@Dgk*6K@)&(Qxjh#3`}iF|Nr2kkdFXdn5BDR?QFLb*
zx}G$EgQXk}oEQefVN3A0YBb*2B;k|yDD;?M1l-;Lr^;h=w@ZHv{rQSq-;Kh#QOSsk
zjl<ic0o-f%oQ4ZM`S(wa&v3fqVd*_8SK~c`9jXX#=!RKs-)O~9?%ds2L29-7^hu8U
zpWL=lbcB}>G^mB#FDugDmFDE@`a|}x-Ji-DcZrqpW>n&QfQFO~X2({IWL3>Y;;=L4
zN$!erwX4B!8v4XhEc>P{+?c_ANZH%TGop<}%H^{iMYnm!<g0ji&1Tv8rQ8|&w3p!R
ze4Z62*9jS0=iqp07)E!Si4i)S51YcB#^I^B-@s>hm)7F>?~MqT&A`dz4D8jOiGw^B
z5qvutuNEhuGCm9?I)#|>o4?ZpHzeoj;pd4s*gTBK+4=f7RbmCn4Q;p^TcFsxmU=(v
zjx3$KWSsSe>iF(;{m-F@nLQ5oX&_#^XmHv*cXjZNz$5Vq^_G{Sf3+JNKbAmACzR68
zU!aI5;SjEVB$r8izVxq>vW}~x&?b&jAFM}H+!zXd+yiA%rbxPZlfDW`6x4q#bPnzl
z+Uq-oPMh<B-aTXHa$X;8+k?a_i*L#7O?Hd@mR5^5?&=FUoJ$gMcMBhR88X|2Nn+>C
zT(&_H$LtZyntzQEUCRuaQtN$If8Bw}J$xxU+-I}o1@|HBpSF*My!cOI{b|40c3fMi
z$y*{bG~ry{xdCY5``x4~nUEY0ML2mtY1u01WX!;S{u|Mcd%6@`$HH=AAP)9Bi2l~D
zXw)c%_KFy++#iIzd--_dGM97Ic)r|#^Gy|0@vkreH7^gMyuUT(^LxKZjz3zGb#U=@
z0EV!kn6PLIdPJ6@NYfZvu8xQ@9f)F&WZDtT8K+hQuzLAIC>kiEcZ(@{Ip|{Kn(tJ;
zsxS0T9-}QSN;r@;fU@0`(er{Hy!*w`^xx;`YvdE*%m__#<-X@La=rN6X9{h+Oq8kh
zSag`*Db%m}Lpywf*el;=3fOpslu8~;c$#0_IC~d4I0lFrBdUZ7hknB2Ba>*dsVd1v
zl(U9|MZ%!V>dduuk|^&~Da>_9V0IOyqTaAOtW~TOQf;%CuS1;dr(-$`I5JgeP)KDu
z#+<{SvI*NS1;hBe2b_k7;K=b(=sfj<!qcs|dolrQqhk=D<c*zMmqUU~-hJ>#?>~g-
zZ-Kb#;EwN?`(ei8i4e56<IwDl2$=nuss;u_@`0f7Y5^L4XyVzH`FJ{18$Z{@!KGjY
ztW<+>a?BZee^(U;K80h|`EZo)-he$0O3=TMhV_Zvu=Z;LhD=(7hnn4S!-;!r8+nfM
zfF=(A(8X03b#$M6opuT9F*<n(z21=`#;($!pVJ?btTIzPtky0uP2)V^P9_?@D5d%{
zKgC8X&ir>D%|4Atpp9P=DEOr&6NkA_m2+RTstge-bZl4y-=8!bIv~1y%M}kV$&sb(
z;oUNsiQqoq29pGJVK)?W#T|dOspe2wb=|mrqR!sef@jo3rZd2qnj5_&6(fw;l?)gB
ztBu9Eq%edBh2nL-8Mti%@&P5N<>x71za)&FwGKIR3K8AJ&s5rcA3fa>FDBTb%bW^C
z>S&`pdnVE&&0&?Eg?rK3(ElliCs(H7Z`m-c;yc;k<YW}%1^GO+#EGdq@3!zK9l2RU
z{bzikIiqGH(@Yaad<L@bi35zB^>MyqE5>TXP|gv~|5#y<uX^(My)FUnJP)aoaE`9k
zy5arj5%}}?Ce3S~gWmByh`Gj7wBiYpreC$*WD-efX}^S8{hL%LeIUd>jbb-ylf|xl
zU+g&I3oTkyAPnN!aJ!F&vOUKnOyzE&IHseF)p;w(YC^Y5s+8<mbFC<3OiyRh8@|=c
z)UGnw@MI<zcHjE2>3?FSzX3aBTF73=MX=6I&qTQi8aBrlxHHFeUzV`sqqxyO0c$Ll
zp@ZkPI*aDuK~*a5@*Py%&Dl_}jz*?!F3tuJ0w>0!Mz#aT`s8rl`)mxg_Qy5V6xi?b
zLGLChqV+Ohm$(K`eETECW-YQ#EXDA@>!CV260ZVQ@;+~0*ba6;$UW`?nZoB?`CVXe
zcsiu}8))s3(-g}0e_BD^Q7h`gyL=2n_WU42H{PYN@`Qe^34YwPhQpc+>a)O#++X&=
zr<G@DuT>3QDej=sL%p%b{S@gQdq^~NpV+y1JkO{eAcdVbXl>hm`lC0U-Zt%)9rw&;
z{&U;OdiXwJK-e2HJLW0dD<{uZeSJw4yR@j!_ZjTr$s+OP@Jd!?x=+-SUKjNamI}TF
zLG*lIJUerVI}P2oGow+Sg7k0=@9?>imS>(2wYw*)&?*)5)_Jk;$j8FS?ol|_Hx>iR
zV$p2q0{Nd|xN&P0lyyDvtw#!sOnjh~=?Y2c2q?Q}qpPX}f5$o?FPvbvl=sNqq`@#t
z6UG~6K&!+b`?wRuch_>{q%1%{up;cvO@K@39DGl8hD%Ks^q8jz>6uB~yFDEi7ypoK
zoF`6Nnc`)E5vJ#!qg7Fg7<F(gzNEy%V@@_TD7rx7!E9t(rU?)EJY;CA8_WXFa&PNv
zO4wOPZc~%V?6#9EVAf{3yVD7VN4C+#179iHtBADMWznKvn)JBqU^Irl76v4|q?&!)
zDcReMa~ihL_@!3N;A#Ss{wWri(qK`uceN1IS|#QW3t@8at*WDDCQw`F8s;dsL}<O3
zExIO5CaoSxf_+dD%NYAeFw;FFk&f1?ZZ+N~j$R(fiVW|H8OH{&eeOB1Il#Hbqhs;D
zGZnPj5mMt+Ozh3~LA>uE_R7V%{wX-=G#+1KqA}In7nP-Ps2d@{g5f#bBfAW$f4S>X
zw+t2i6FBE{E|M>CA9>{{(A+$ni|6mBOmoie=J(2dqu|IL@Mc0Fj2}<K=-1=0aCa_H
zFa<7(12E;04jSS-u=;@k8gk?@<+&Y-Cp6LX^?jf_wHNGr$01{45k0>#k5X2}B4Lv#
zl=FU6>)Pj3Q>Kk8$0O+R>{9BKRMJ_OJCtaxOjVUkxK*1#0n{R@%}}7oHe;c$<q)=d
z?OrlU+ASWw!9-u%2$o^Df!yvklH(C&3hHLdx-{35-R2!Kqdod$yHttQg>@AZ-VI~V
zV(+nllqF(7&k-zhSWi~6brahYUn|x~tz`PqTbbO2GVzCjBX&C5VtREvet6{Lnpy#b
zDtDY|bO8Mq1!aq=$Uia^djI%d>9hk%k456f{Bbaumkrxt8F)>JxRjWLUO#)n)3+F{
z?=`Xe?JNX^g(0WsO!P?NpVy}aSoqBpt`#9jO|gPfq&qtKTwv2F3wW;_3(sy}>F)`1
zTyx38g&Lj*<QbMze19<E6W?*4I7F}S=FqS2so2`>6ulgrPx1@ONl?_qkYB@i{%J7v
zTh8d+qa2#-)Ig>dDhO0a5fjpSP{=bwnV!{Y?v}EnuiLHJ$o12h<Ll2d+2v|lyn<)=
zodlaP@dqjLm;$q~=N{0CZ(@z*0a?iI&yxHV4MOCGFXD^M<0*x6dlq?2<y@*6tY%QQ
zjmq$BA=Ea3!WgK?t5EDHR25s=zX)#jlUd3IBf+lkQj|-55%<0TyVq{V-11eR8QZY$
zPZ|zgip0l$iEzFbjzJSQ;Thk>ZCkXM&*s+Rw~aTn25Cbo^~238snE!^#}1yY3%Hno
z7k6$DUyZ=8-)Ov;rjC|2Q{)_a$=?lLsGrdXIg>T8$W9J*4|=20Sq`<FJzF{CHks*e
z#bl?GoLysqz#85ScF@9{Sx#7T;tYj-X{6)s8T{L}i)`(=SNZQS+EHCjb^P70n70_d
zg?-r{(^*tsWrxi8QTWvD8U4*4$z6gH;q1#gO4I3q8pC&zqhGGjCyiqi8TDQkV$A#h
zgIuXa{}e6aGdsN>Y7*&cdC~L!ctMNug?r~uvYHb|MJwGW<YK780-uMo3A<C6wD(m>
z{+lJ@wd@Sm>6I#eZLbp44}6u$-Pyz(M)SuAcN;7jSBN2epSJ4oc97tOESEwIUmTDB
zuI@qWHlAI)%d>Y65>Y(E5#{R~U?xe#^!};1>Kcx&DM1+dc>#>}xguNag)uz)y*i^8
zswb7fcI!a6>Doa{*B?<XDp1=w1@ANXPQF@&@8`L%ulFN5Y0r7-zphikB@;}zJr0fO
z{25vLje6Ws!7T@0>g_uk&l9f_jhB&s@&&RV|DN)jPLRvtcl4&uMeadPpo|~3!j3^Z
z1^1bpvkMhrDDOko)I7FHpRCI=XDCwSyd0TU6GZRer{WERb3(Cp80jsDChzjqENFNP
zJ9tXMisj^m>BoDpfxM?*6~02a?Ru1Heez+KhwWf5e)+P)8SS!B)>{}I$`J<?7Ez6r
zg6u}SCF_V=FL=$J%&siV6AurWLe)?9vi$}zxR?<Or-Q+$9h(aIUx6so<N5l-*=S3c
zidAWz&>FA?Yr9OuqBq<tb2%Rqb`L>;{Z^c;^T)R%JR9D649Pq{HR04&tdH!9%T=my
ztL=&CynXn~oktcS-ncW3XB@wD(44#kOz&%h<)wVD7{8o5<-Bmrxd(K{D`P)r{Il*=
z<jCL2(x*Iww%v#CuXd31e0R8|U!#m>Efj9^#YN+PBo`HD(;?q1vj|TT@3#k1cHL^u
z8`8!Q?nv|Q#&`GucZFi^B}i#LAe6p1MoNphb7R+FDl0uGX#On~-JJi3gHuN^?Vfxj
zV|hi=eDJJH$$GQs*Rfq}?Nuv;+<7T;ou?$ktWRJq0g6oT;ta7(dz}#Z)lINFw@dUb
zbQ0YcJP`xV_Yp6hn8Wfl|B3b^qF8wMnXGqa8m@A7d}xF_8u&BII4_K!O_OktpV_x1
zx#QaP6wZ&IjNu=7e{{-fXiUlAc?~{iHg`t!kMWrOwI9CnJ)hn6IM@b-Vc)3)OpV}s
z^8^7l|LsLTlOS}l83QS{<6Mtm*yw7)I<zNzf3=e9kVJSiT%k73Oi$iqg>iv{(KVZA
zfaEJlKW-b{TN^{AqknUEb`ONqJf#~kP4qDCj}Q{V9lK**(4!At(2Y4jNo$58JePCh
zr#~mN#`&U0XF0X=981YRS3FkUPGcX9V@{g+WNv;z@Juum)7ST9A&boD;VJHp+;l)z
zo_~gt+=9uje-EZ@rp;Pg6j}73lVXIyWHQRp5DzJ3v8)*nWxLmAF`d?YQLnCEa>a-<
zfs8oY_4XXr{B#i0`+iK~TGEf?<QuBJiul=7XCo9hWniK1N(3L+fCTeAm?II&GhFbg
z#TWOcyJ6&zl`x#^h7Bzq(CD`Yl{dHZo!}JsJ)DZ6+knc-aGdbjj5*sD@VSTyZpe9}
zbs~3AZ|3`_-izTsXcjUO?4VKNileh6kX`#iW@iTD;U_or%2kE=@vn4W%MZy7YVg!I
zLZiiVdR#n-4llFECHEWD?|C|R&`2qu&wIgVt_ma$QqoI)Pelu_3S+Yq$-C?uZ5rWD
z33nOw?>s`fS)0kq-IA3n^Y?q`8CiJ3D<Qgetso~}sgin~k+fa?My((7srlwH*^nd8
zDLVC+XxBDD^7i)yQSQG#HgXRVMKepO=q7QXTLChPZfy~2w#w7fju8}jYqij{$(c>6
zNoR`gne0Mcni$im%YGW?GN1GOb1zRr`cOY;S9oBaHTUb<Cga-Oa1_+9#DQy3kjZDD
znfr6Thq=LijubOv)<UIA5bovj=ipr_D!5y2R@_E3r3}QHx6$}}dp}C2tUzhv2&m7W
z3y<Jf?tXku9bH^e)MYzL1|{RI&v2M;2}d-~SP$*1fTMO3(e0B7YT|;ie+$p&2A?3m
zo$X}driHYd|EPVC5?YUR()kUCsOqmgz70zvxkJ0ib3XSWCq5E92L|A-uNlpJ+lz8E
ztpumQc8cmuBiH+zX@B@Z8aT3u<TO5$*2jT@(T66PIKL<BIiU}8{h%XA)3-}T>bw=@
zlsl?AY#`*>hO$Z4%~YpySk^RILD2dU!C2`Lq4|<j@apNw>KsQ4w}vcd@9Zy9G2aV#
zuI10@hj+xfSK6$-xG!r<&X<j!JO|_RH^XvvA|}nUK;>31?Bb3SgAMcWn7d!+Iu^o*
zXVHSMtcQ7SI<m%RK-Yoyad`(IdR!R#o(RY2kUi*oeI9osIl=rX?*TT?LD?&Bq^+BQ
zycwoYz8Z@lKKm=(VugmePwBpwC49?W@Y=_M_ng!*Ur`Ybdvs87^({46+@&io0&&h^
zJiOGu&|3v#jEYgf3cqCXPy9|l)gRE?=e#2}!;wsr=TXCqTGDeGjAls{)qd)W@{1DE
z{F6e)d#i=!+nH2zw2AYut!;{x)(d*=5;nS1p2;pwBkAWV$>O8usng)OXtr{=*l;b9
ziu#-o+sb#!I?r7fnqo$?uL}+dc7b-X+9@e4>xZfs-f={<ZjxvDUk3<w$NCEG!%SG(
zeSzFR)C*O;JH))d;cU^GJSJQ#g!^Pq{Mb7I;)M{j>dc4N^DNxpGvS4&(s1<hR$Ox2
z$UBZR@&3_Hj5D2thg&A%_b4BH2#W@_^4Y5M5?pyU5)KWMaOOt>_9rFc<VX`7>92)G
z&XZX+UXMEu#zCI%S?$!mkao0;G-GXX?Huu){%<;PcnU@hYb3iBpUG-`Eme<OfLhic
zgC4Zd_M=;fhX0}0Pv4Pe>>A9|+)OV&pQW0}Y&sC3hfO92Nns0jZ|kH})MW?S>7h%H
zrr1!sNxI-<byFxCdO+f5BVp&dZ>Pp<YUCHw#yNA^Lg2{3bi@9TP?w?4jFMEu$0It#
z!`#WUWW+9d5wnN|sE-o8W=&)PHw0!j_ol?L?^LGuTVL{iwkwkh@V1t#dwfzdr8oUk
zN@Q<sm8rPZQDRqmRIv8Q6?d!6#NtvZq(z(Y$ua|1yt1Lykd799HagfLMb%q7xPGzY
zyOmH}o0*M#<7m8omWUg?gVPyNhB;YjX!dacvvI@s1K|j4H%31x&oy5dh47(Eq2rf<
zJ@s{TIYk3@hb__L-BesolE($^Apd8|U1`T0ace_&jC$1%tFD-0v}!E2E04ra%Mkqf
z{gm9)22lvl#EsCa<bQ`lRHxTQN_Vy3n)!{?^DfZ$xDVuD(j8tg8-<D<{m4dsKGn}k
zqZHNc!rKUAYMygjjJCc`%MWj&Vsk^8_4rs(&Ob+%^PP949`q64N(I67_+g3V^=p)J
z&0hH4uE84Qok-oXLzdO0STg#FK(a0p=29_)sjFI&QSN)uQ6YgpbEonSMT&6iP6(3@
zwXW`;w}}<MUN5mL=&*SxY(dzil{gp>3$H%uI2_`Ph0>jv2VeZSkb>iXEpYgl5xV9r
zM?%9Q7#jpZbKWv2@f^BRKrwz)+j8HwKQ`-6!L`MFcKtIPEo6-@+jQai*B4ElCfK&L
zjcT^U;eKQ=N*kil%~lJ_{k73?*#ypKy>R^H5C~lqv1On(zW1r8tj4<(=Uz+0c3hwX
z!KLJNy(jMc<-V@X$~c~~o;vPM5&o!bAQ!nRs!v!)X}2B-78hfwevk@6GVKJTpQFWV
zPHQRQTd5#@mnX=t97(R1=8JazO_H|#?qY(qE4%VhhsE4AB46u~f@A;r;$fo_8f(3Y
zp8gJ{s;&_1W7m+Zu@@!Wk%@KzyM#;TLn$WDgxUYzNjHT!(t5x>b!o@N){em}fAK~t
z&GX~TUo%$mwng-uHV-M=H)GkC6<D1Whe1|Lp=$086~24g@+BO$7lR<LxgF;xMPWm5
z0>-t+;Oj7V{8|%$q%6+kIOd8oS=l^;EMV?iE%e!9i&eg+@aCQHuH)uI@q15vI7V3d
zZX9l8^~0rm-B8Z`Cliiq;H}zVTuIV}ZSF7XSGEC-y)V<=><HXRRm7m=B~a%qkEi=p
z@pL%PZ10?i2%7;2O3o#vImR&7iK7}jzT=<SL3KMk=rlivzkQx4n7pW<Jt-?Gd~lYK
zab*HIE>uC$I|JIWd<@$#&=_(n_Sqc1;4L(7J0|)bj;EkPRkpo0k(Tm%oqM-v(rPag
z<k~M(d)|$u*6jXdXLnEV^|leKOr6Ez!gXTww%wv{Tp!7b-A=4sE1ljqu48tO8miS@
z?uvCi3|U9%T2V|5LW<UEEa5r5nIGNZc-{e{7bPLidJBB`w>1wANUZ^eD-3{T(KM(y
z&qV1(VCD&Xd<qGJXO0*49|=PG5`xQ3-g`Edpz`M=e0%tfTzO|H?7kT~o(_e@3k7)Z
z9SP^S5%_4q`}MXz=zjVX7!~)&=73%pZ88DZj%AQe*WYwT*#swg`_Q|+Z8ZF?DX#s?
zq;l@r3+UMeE}4hPS|^nP^uAE<4qc3XX^K|G(R8+}5+06J#!^b6rp?cUDw70p{{r5{
zo^D}NJhMeMv}c3h>qpec|IZHpHzXIQDVyAX8S~h4ltQkpBe}+h;-b*8)TZ`QlHtBj
zeBDQvRn2%NG<B%6Pu>&h{Y!1mRaO^rwhX5L*N<X|qbdsskO}R!#A=#qW#zs5GegO0
zR;DK}c<-Ny>~RS=pqvABA4fdxlZfa{Yo6ogem$Of*uQ@>+Kc@0z0MJrtzz-|j}vw$
zFM`P@GxQv{3{F$Rcn8`Bhc=jC!Ov+h;!eape6Drq)pXptHw%?~x4I|3oze;)lEtv8
zIJk)4@yEB)NmDPJD7Hb)EnCEd8sm<79F$sz!%X^-w6FC*3BOx?>9>>i9@4|+gnG(x
z5)h*-5C51uWRP^3B$ho8qfo=2sYX<M>Z))ht6Dgcr-0Y{3MkU)s~|tJRA}~alkGWb
zM-!)Rp~pTqger-)ShXia>~(4#*&Pm()P;PN6&EB6lD6K=<=Qq;&OXdKt8sxKryL;B
zD(^?q_9Nuzc}(~<cqG}+8N;rQWAwPgjj7~lvZ{H3VoiHywViH<%qlU8)I%qVW<PDH
zx}|{S<>jC)HwY7>!g05H4m#!qL49U8dK+#><|Zi~Z8k^wpM?lL<A=k01JSr_Dcp)f
zF+O7*PT!Ma@_;~u4zWkU?bVo=l>*bQvvFd4JU&lWK^*6OEZi~ya?!5Pma4!xfY+MG
zMB;wCCtM3%@hNyBrY!$S*Ub81`{*-dxwL~SCz_!k<RkgE^+8FGTI%{fiGB||%==^R
zq?Q^_9g5u3S2d1q>GAu>nHTiyZ!a|M@}sOlPlS;gL@DXnf}G`)YK=c()OuwPncvw&
zUn|4N!uYqqvZhedQEU3IPa#SB`OEzE`{D7Jt77Z6pTdHi_f%uBM3hV!M}FPcvxI@|
zf^M{gtr(+5O&za9i)k4`o1c^5c(jjL_HvBqn`$f;YnckZ*Zm}U*QT@L-*W`r-JUEU
zbQ4QEWeL4M(J=4777i&naKTvkUt5Wt&5qc?&+J+!H)G1AO!R7;1C22Rxx!SW^1Si#
zg`6$&yf+H?9*`X+<e!;|q>mcVl#IkV{`rp4<L{fI@mSgPhq4>)(NLp>s98XGI=h9w
z|L{a`eHP^TbJt;2AB^}l6ax}<F@O6QJe^lX|LvZQVVijW*}xPF^gfXMjAOJaGJ)i`
zt)!4&x9G~RW+CCfxiqDrnmo$vC@OD1cfHOcb(awoH{Fny9*L(4*9U_9%Vc^mayz-&
zl}g%DdkfYs`GTJ67fDCiC8{Z|wrTM0$sTSFpr(56wun~~4|DdMy<sgadfA^@*fw(9
z;RCU1qcM2}PNN3*BB9x-R#d5~5!GFbB{kPo*x?He5_gMUlr<rgbu2Jok$1Z=qee^N
zk53?|&mzp9v>A1mL$Nk*CC2*4V=K?($G_n-)B~Qd<4lrOI`J4`<PUdY7NSx@VU_9)
z>(Bi#Qhy|}Bx6vbI2n(4&&EL^4nw+5LDqW*9P_usjZcA?HFXheuKq{&+jODUe+Kt^
zFTpkWHj-Ot3704X+^+70TF%<aHB!cvo}cLx|1B)4IYnw(Kgn7xpEK8I<7LzZlCB$p
zP@Xrb4;_mtnLq)-S+wBuKxn<)B-|Oqx!-S}P{=&)u_)Oi+RfDE8R`ge&51^y_1h*J
z_TMh@H8T?A_My7$$Q>a`Nt=B9CDbz8flB{HQKgVTn@8=Ykj^ln&FQd2ZhNNn%)+OX
zQWqjlU*OC_0@SFkAWm|)YpfVKZI@WLOPw+-Zi{=8*E5Hv8)8cCI`J^Bi;-?WMV$|G
z(clsct3T6_D<_5e8P44rv=&jds~~N0!rV(d?~s`R!}b+;f5r=gZ6>4iMk3^nI$-Vu
zN7VMn!{x>>EaT_WTfa4+a&;6|j+zYH5&XM1l%M?_2f*p+R7Bt4yd0G=c#(OWYR>dR
z#VmKY6joF8BO|QLHO6TEyEmTYg`!?h>0r-`WZIyIm<sOdn*WwE!?I~>uXB`;)g3oi
zJfyJGeew1BKf1-a{}G=fsaKdhYEOGp_L-4X^1_AQ_v;BK*HSWnV?`;NKj`N3s}z!W
zOOk(mJ*nR{6gytza>k?yDJ?1>Be@7M|6r#ub-hTFV^v9dI=NbE-Xe>v596-LEU~7*
zLv%cHQD|MACb}({z?x5glxXq2iDQ>Z%ua1zb?xe&^u9&Hnt%5c9hVVn$(hIWdPSk(
z%>?|dNkZoz6&OcMgVZM(LBH1_U_=7y5r-X>NqC~*i~3mZhm8n8&OJ-m@%?+&hIqt8
z@q91OCWUKwqBu4P?%^iz>+6MyL$yE?`PpmLG?aVt`}*$%I5pb__M#d}ehq+Pl_vT-
zEycQE?(kdsi=J9q;M(qP@RBbh`euQ5U0-mIv>y7c=#O@TI{Nc6j<#QqqZXSs3i5Ks
zotOGBelrBxnn9HBcY^Gv1d*MsD%Batk#xZ3>Ol>=sX|gB=y^_{gf*fNsq$PX8=)#D
z1x}&|MFU|Sm?FBWl#%z)Iw8PUCV0N|CBMKAqPZVu*AM?HYc|!B<Xe>pY1TuS$)Xan
zGk-2o|JOsPvwtrBJWxQjy!)S@ph3QTk1Ac|VAImI8%roSAT(c_B7X4?gwIu9cq@3|
zeWV?l21~KlD3E_H?wHQIkb!%a<NFEjC+GWnV@v*<pEwx(^P=(Vq9=UQ-7q-T1<{EU
zu*1z6!KyLnyrqCXb(~xLDg;?eJfOe5H-^9BK0kd6&gGbg$sYU;aH5s2cYi~L8J9?F
z-)MYn=!0&HC*bu>ZCF<ghDD(~-ei<e$ds?-+y4e_`}LPzxZNYI<M!yk>nfE+ap&N!
za_W(IneHDyKziX#)X?@?(6$FT=*#0>!8IbUv!s*zm~_3mlXPDKRcM?dGn=cD%z+YW
zOPM4#ac-(z{wl%qu0%+gm?H#C^ALPF$G3gYW-8fdLJ#e`F}>1HVx8M~GCQ?SsA)<Q
zPrh?va{K>B(Rs&n^?z~PmYJ1rL{>wQM2XKm2O%U$ND`7%T2e_WitOwtQbt*YBq@CE
z(Nsy=lF}YhskF59d;k9A;UD)t?t7ebUeBj5=sMI2=)eRLhQ!d}#!<Xj!B`55Rpqm4
z965)v`ZQf{ChIeMpw{4Aym`$VpWa%BAzJZxz0Dq@_V2<~FIivW_$ri&UyFWI9Pt<9
z9>m=;$L!%#(VXRPr7l|Gqm9e(>9%A%8t;Z7@>|e$-aNb?z!+)MTv4%I6n}a;;BOHf
z+^!OU>+L4vDz;~5i5OtXx09e_xD1V}bg^X9MBJM<obki1Lw=4js+}8-jT7I)j^)R}
zMfE83FfTWBHo~GSLJ)2cN3R+e{C6@FN;9e<*!Vi6waI|@{C_Z6D-WuxVj)H%58OBT
zaP!ogAo*+*H?jsmbOjG;&eu70muR@Tyj|#Ux1LiNQw9g#{NhF8&hpM?PF&@!U*I?A
zJtr`fB|~l|-(<!7G|gk7#5sV|!uy=a$dSCyA>_pdEJ@cpQWzzDh#%JWg|Dlh$k|SH
z;cFN-RdKo>zJ1Nk?r*2zpLM?I+Pno9-CT;nW{D_BPs9pUA9MypB|8foU+RLLcP3)n
z2?55R_Cc*(wph&0+T$YU;D@I5_&a}qaW*aRZ}DcdigUoSEovy(Wr6E=>LSGn(EpGE
zR<!%0#}$2yY4yY!r)jL`eH6Ox8nU~<6rA*LD)!GX$L4M9Ub~XR&I>=G$l)NgKCr{r
z|K{MjK5={-HVo(I<-<&sd!V`ECDe3(1C=l<5EM6qc-jkCPA8!*z>kyP=nNxYz2Y3r
z6`*9kyue5P80Y8M$>k{?0PBa<aO#{aZ5q7?E(t<l_clkEG5IQp>=$ybK?nGfyR|~W
z<~7hKw4|?{`?$U(>p>(tkTbh(N<P!-`Q9rtKzG+IVXJTkHJOc}>`N*1IY}Nuz9sP@
z897!CZ4)?aD|7CrjvW_#YbKYn*_ZoyJcSb$gmR8eIh<H)0_QBbpBs?d!(|Tc;A*4x
za#`!PaRQMf&eMS7RL>c3`TDyBVP?otd;s@LceKD(jRYd%U5#ZSYlOcPc%jHY4Xgf1
zBl$4<6st&aLt%_}8gKjB%gSqExA0E#QhrTYlCUeXj4v|o<WpyhP=Rp-Uu4kFOE~CI
zPWM-S+<Zs!5_-|9m}!KgUFdqtPKr5@L^9nuq`~iDJ^B;W8C6LeURKjZ#mgk3-bOOt
z+sLcu0=>5Dre>S3G+*r#ty(SK@~K(0<>I{Gv{&jq3DjTFNvY4In)sf=9)F>t!ftZ0
z`cC_Pb&^5*Q^qy9OI&0R1wOt)l9hKU`rczw`hJ_t&pe_t?LFjR^^{z`_fm!0M>>@D
zftrgxQT@gr^v7!#SD(F;3xs&iSvrTacw59p2Q+aXVp(0st&@AKP|QtO+Q>PuI``pv
zR*&%N<<xB7a1Af>ITepe&cSFO7a_izqm3K5)6+z_S>Gmb7V(<}A|re`kvroW<wrdi
z)Gaa-oE2>oxYwN&#5}POsBDiHdWGB(I2ha(=I4Y9V$zoiI{U8+`^P)+zjpk`8+?r6
zkA8c_M}N4(%TBw^$Ct>G$zmx&uUL`_MQYLarWb!!Qsk!PB&L3h-j1juw^e(|=jQ>c
zJy%YTGFs_s!8JP8*1__l*GS^`1u7VOk0d@nqbnDG)6lXZvhN)r?W#`NwEsO#5PwMN
znGb2!<%{&f?>0r|Tp=U1Gt?vBMW3g&P{*Yfl9<v#v+lQ(weoASe0QHNDSo2m@h>U-
z>{F6b{76+J-ciiPcT{KciRuJrxr~!SZf0R4cebAAG;cR>lTt2o&ay3BTt_E&LG3Lk
zSaOEjEIq*4UpmA2wW>ju?^W*ileb*cE(z$oaFeSj6$7Ii?>J+(SKI^b0gkiW!U-3<
za~syR3ffL>=b*iUQ^I6H-G?j9`tzNcztf28?fZ{w)h`t!<j&#4wrv(T$o2|iLbR>g
zX07EFK7Qs?N_%-z$#+7nRnzDIV>wPVE#jAXoa47oT}TTTr_qXe`)DB3o4h=?Qkh6O
z@lWFD(bX8*vm}MKmz0oP_F)n+y-3AfPwDZ(2J$a^O9NfEDJSk4)yZ9^W964=$hnKQ
z9=l8nBG1#WrI+dcnA^1S%_)*heL=<jZ)o|q`(&+sg86)IQe@k8di?4<9Xok}ELYql
z+t?el-0C$gtGq+o`(Ke0d!89%U(>0NeRSLECO0DR7iT$mlM8G)&zZfJhU=Ha;a~R$
zPE6tvcW}{TuKkuMM5>Fxlv-KvpDPBkY3I2kOGm@mu@AY!79(Nprcy4-UmkA#{ltZh
zW$r!weonltgR|TDh1+%}kK6a<6{nrn!U+^Fa)uYoIW6W8e|RUysv;?p+gZr!9hW{@
zIk#4F&;9bb81Zw>4!;99C1)MsnviVaxZ)lB&Ucl3SV)hsW8GK&^MMO|xB=T+Z|RdT
z>i}IbilO*{gCwD_fiAo%q_w37NLh;IaxSKkp6CTKkUvk)<0|Q6UIjhKd_z*BZ_+W-
zR$6Ovk$er#lcPx=J;z7X7+OoVi>{LC>`zpbUQ32QYbhf^NOP;Z$TX#z#CFxv;L6ja
z8*`UD(_5%}V+Y-w{f>_Ryh9m94`}AD=QPTshyFYJfj(*e<of@JLD@M;aEO%y)tOST
z<H9ts9@K$d$>vaeT>`o$Yr(N4qhMHz5_tX_1`{=mq5Yx}$S7*SmSQtl!`p#G^h9WX
zJPEqD$-?OI;$YD}$VI-_h6=_HJGbHwSG!CBmIw53>1R%J3kz$wlp^N!j6KIi@*3=n
zSH&qlso_MpTt2fSgI|6%j?2F$!MX0Q7Klh>SV&wG;T#TJ<aJAunnPzAQAb4tzriS-
zKfQwZhQphv+g^*@pNymx1x{q`??I)->qvWT1*v^+B9D;;w0%}N$%W<7V*PzI&-y1B
zHFT5CWg)G%XeMcetIS(iO0#26(JW^nc|1Q%O;HEg8e2{!+O>2tpo<Iz4Ww!Nm=1q@
zLA|?f(Zdz>bebAzc;sVpcz2E*Bc9Ths0Sqbri<iHd?v-U+EBk#4Y~sVgBd=SpdUI9
zs-8%LrI;7!G-$%AE-Q$dq5@yPAq=l{gqZp3V628C^D@Z8GhYR$x3-7c)alS=`j<2O
z8VtEp?LeSA0S-lp!SNF=5EHBqr%p`(%f``Q@<JI_)hd9rewo1IswC)1IKlQ0+K>|P
zo6}IVfC0DGX0vfs-1&1U+{1h0xDw-Pfo{)3E6U5_MxR!s&W=1`Nn?YRd`TKVW~Uh6
zpfQ8*@)=J9Cl%=Qm1HtY(<M%cqZiY|>5^DHjVjNi$nkrq?C~mE_a~R!Hr*mY<R03!
z<0|E--=UpJIixxB5G`F&Ks5`QC#pJ&^g1uos>3Dp+2|ZC8@Hb_TP{$F%4HINUPc+x
z&E#Nvm7Zpur#78e6jgVI&S!j~MGbeTYFRg1uWyl@Nhg^l&H?e~<Dli07DV-qhYMf^
zTcxLgBrSnrt3Z%iFcWN?>_I|u1+2S01@>mHfcw{%Li`C6xbY|+>`(86iak4_AvzrH
zno2=dlL>TfmIO<^WXMlG#2M>`gZq3j=y#q6hAZOW@@fZ|I!6F{>;CUTT+CJIt^?N%
zc97wQERUxI={ldeUsI$xO-$o*wWByOe<?6@H06}8$qGAXeHHAS`kgoK(&3&h)+X_{
zY*t*YR~Vij#D86FOJNm-w0CPB`33axzj`a_T;>d#?Y5X229}ZCokwK;A&5D5c2oQP
zP`X@RO`e<f(8Jk8{<jWNM#Fv@qf<{lZ2g>EdWR-P9ixBVhiS1%6<vwzp%TX6-XmE^
zzvewB&+|t}c&vuxBkJhuy-PG_myp)6^*Ok?lUi3jBbTggu=Cb<NVv@kwqspjVXQMq
ztwdO!>jCjI7eb1RHLRMI2*MfRknR%-VN!F!^Pnq)=&yjWw^l&V`BZ3gn+t=V7Q^*M
z7pQva3x|9E1J|Hf*ijq~_Rn`ig57D(RY3;EY+!kri!ltpW&jQ1AGwC0A@1e&+1$Gm
zi{MxDMzB67fK8rDz~#O<RR3z{a_>(7b8Q8FM%F|~Uv`)~e@dF=Cr=5ygkfAj?Q333
zHI$3k(8WL4IF9^WFZ2EF&J=lTG1-p1&F?Y~rcELxbli3deHNFeg=?b;R~ORw#0m<D
z51=Oob+ln=3Jtu8qdhZrlB4ob5=LI2peYAvvUw$yFD)isyN2oyCs9Cn3GJ0XMwgy8
z(ptB}wEImRb@ZH|6HPS~lh#Sf0sUmVzk?>;yg=^qS7@Nm3%V~XhLAP$;lLsX7!f)H
zW_<I8YkRU__mNz1Q%->VEy>VW6b9|j7lVFPJT&Xa!lQ{hK}FFQ3MA6NcHTIMF4_Tx
z^Vs~zr2??uQw3rwMUZ)L3Dmf)fJ;`-xixdjKuXdX9B){Ig<>Fle&Gv_>y_c<w-R7j
zT3B>21E#j;LCyj%Sa;kS+TQ5E@u-bpJiM7Rn-aryAHK$=9lgK>y>8{ol->zmFaO5v
zG;!fn=7>;0@+eMY>KA^--|ZCkfVbLYaFQ=^A5QvG8NB`hSNd@DJ3l4IgEqabp`y_S
zq~tz>f`5;pO#z3g{>eoe3{R(|_yW3Dzn7eHu2HWR<CyO|MGiWbDRM_RZCIQ_!=+Bp
z??)9hWLZRO9PZH6MHk5?=OXp|=%Oj<O>}DEI|_KmIEbQWseB*%-(!<tihU|X8EpY-
ziUH65L-5BW17sNcr!O!Y@|r8)<nMzp5OEZ$&Zog557q~kX@XOJCGfm#8(7w@Wc8t4
zP~V;m4Hq3iFFP8BtDb>}mrt^F^#C{yxPt!RPAD=t4warKA@Sup*gJd$)V)`RE%PFQ
z9!5jPKpk|{Z-kNWbRaJE6Zg1u4&-L<0=3Lk5S$qe+e$(qdN2sq<+B=oyCWak8O|+#
z^@{f_+QL8U?BjjsZRTZLRCy8i=`=yvoRogw<A=K1`5(hq^9xGi$Sh(9Uv6wnwYQvU
ziPkHc>h4Q3%Oxq9vA4Qs&7f;56G=;bH+>0SPo}>QlE#%i6n?yllI0H4$=Z{&?9K^V
zb~1-1<s2lBDQQ%yd4s|-^5~ac8*TJCLW*q7{XE!7$);B*cWWek84&~?H(bGTH>-tx
zPXQ&DRd7jeD~uE#1nq5y;PIafn2@pwAbTf7C^tj+(RS#%vI|0<2_ZzH1hhvThNH6&
zfph3RP&BNCy5=SrUEc~`JFmg1p;M6M^B8m&ro%MhX^__r0LzUz;4u6=Bvlte?d^kb
z=~WukoIC~hoJ!#Rjy#Y*yd1*REa9X^41_*ly}Y0MVL);m)Ko8l*yr(ZGgKd%uRY^J
z7TF4uBcnK56M?{ZVGgf*{uLi5v?XVS=`?X?EAN^wLLNQm`NQ|834J#n=WWA%X?{u_
zb&6l%ColcTpSfj22EHdKeXleHPw=JC#00W23nbg2P%7D1K(mkTp>eN|(a4Gvs*&45
z4N=wP)1N?E3)87caVPnGsG!yH%+Hm0h~#!xkw@qu>Yr3Ypa1UwQ7(pM!HM9qpb(xd
z&xIvVj=&4qMyQ}tIMP-D=Y}7IlIB$Kl->vHZ&bp^X`K*1<|5=(K7b=?r$IdB0?fX$
z9ZDjZGvz#UCBAzC4j(?iv0Z7fpr#tiY@e|@XbYr_zXc5e-5{oW7QV)AgCxr{pxyce
zra=M7e~o0koqcdRxC-3XtYx0+7_ghx1mfFM;r6;B*ek8iDfZceX6**9$m1Kg{go#K
zXiNu@%DGl&b<?;jUq?`^>1!_8;H1D|WiB7OqeHlAfj#ehP=i;GzR#!6mf$BTKIUgU
z)TWWym-xFPX8h(w;*?<<O-5@=`0g2(c+(x7yir^MEprhgx0F!&{jQPBuK80>Q8`5(
zDWlxn0_w|2r}mC!`tN5cx%Pgevw2D6Ci{vGL^RXLqh%yuu3lS-Ovq17greg)pygTy
zBb%yWJ+YaWOS|Fg{$kkUbPm)9PQb&T9T0Nl5^R#c3x3UqAZgbT5G%d_TC0m#Pi+(Q
zYvjUDZXML$=>^j*Pr&N!O;GKXz}o$vL8@U8e!SfeSJaNeNQ)tG6!`{i@ApIZ%qLv$
ztRC*AAQSTX3n4@{27C|xf?*%J;9Tke+&Qoc>TOT6xx7ZOx9bq7yf6dtsUNs8ddg7k
zoX5wXm4iddyP)mM9AU;sNf<b>2C~{@1mRbmpv}(|#((_At@>EWm9EgFRWI)Hoi7Un
zT(LGKP90CBN|Wf4>nY0FU_h<>Extf}9CdiRQ2wMsb~lnGvF!(_!(cxZCgf4c>oyun
zJ56R=_S3VMVI=9~Nrz7tk;tD~dRTRy;+wxva^*hiJamJG35#jQf)3amRtZ;G?)~Me
z28g{$Fu(2y)O9|E6YD#`c<mcV?s^S3Vp}1y;Us*W_zv`5)xi(`Fi0tWh8v-M@MXgV
z5Db3`ci23WZ7cJM#0_UNkNcrnoaM@YiepiJIc#08g!@i0U-Zt~u)Ikg_ul>kF4o;(
zIBF2A=YL?%vIGe84Ts%(rSZ{p2^{@d67P14WBZNooSow{c>Qq+e1H80wxvIZ!9Qv6
zL-H6GDd!BiyLs;QY}V6Hm4X2Y0gQfnS@_^W6ttC#F&2zD93C}AcsD<mn^<1RPigR?
zw>uWms?;C+<7ZJ+VSkQay}494vu`Bt;j)VNEmkGDr97p5Jxct?xfE)jPp!Xv$<0BH
zn(x+=cHS7;Et5vuw>47YuvEG?vXD}Qcd6vqcRKW{jOuIlQm#q`>5pam_VGOE*?J0e
zR$qfj9y}bl*8^wLE&_i>1P{DD1<J0cp#JYmP^jwx7l~fT_Iw0UlRm@Fd+T9#?=Wn?
zc?!M+zJba~qNp3y1xvKmaO&anVA=2uwDm8+!&fR8Yomm(dtSiR;3iNyUI%&&%IH~6
zpqi_N*CzY~o3;{&4^~6F3B}N%)dSlfUtv9wc$jOb3uot5Lba3vh<!?cIprzPpOnrG
z{&x}n*iIlz^$57G$XvE}PQr}o<sdE{2QOv}18vJ?aP8tnzFUT!wVROB*s@y~^P||(
zP`a%7&I>y_wDS~qd1fXjzBz^;3ecf8g$2~ExrYyJ9pwF{ZRW4U#PO-+&a^dk8|gpZ
zO@Xgh(QqR}+WY1>)fJ}G>92Rl5suK)@)Kmrg_5M>d5YeZLB`8ks4wU=_$d}Zf5Ihr
zW7h$j?%#zN#%Of@cLwx&t03F_DtyTm!EXanSYUPrYMXg*sV)NLMFi@OEwF|whAh)O
zP}dlSCZFXnNc{l}>7>C)|0l5c%Lh=4sQ?W<e<+_>45|jVp#6vrZaFp#Qy3>=g`^Uy
zX@x*|mH|HBPzoN7?45mh1A#k+<3|r&Hv9VlCXJQHVE-#z-uJcOdR-aU7_xe&N)RMo
zy9JsDmBGr^9#%DqfYGI3_-hi#{e8e(%QgE!$>ue;qo6?;JT47>8zsW@8?#v^WXQrm
z`V61OdP>4bQNe*j`T~og;dJ+@0WDuq$3x*$zI>UGj4W=`#F>%&%XezjQyERopO=xs
zwIkGIGKXGjW|M1f4`K2yN-26u?gx(0s$S-gn=n8V*X^f*{~i$SxB)g+XTf%SIXErq
zfx;I9;H&Wz{Qf?G2<?M#J3<Jd0ry}&^9AJ|SI31-m*DYoS-f?m8qSRq!>g-*!1>*m
z;drGIUS&R#wMNqTX_hpu#j&U<_YPLy9fHCil6bwg2J}pg&`@6jkIc2eb6Tp{=kgoQ
z>xrNbqq1VV9J&V#!26N~u;z*~K9`U~DH|tf@*0n=<KBRY_Z5g5H3B8JbhE5ZCDi?Q
z3v3MELco#t@b23H2=+gMaN!-OS!e@OSUq{%{5;Oot_jjK5<zWBqE)rD4k?MB<2O$|
z$t`}LMrwO>;qJGI{N`Qa@aRW8-+yBg?bV)0Qi}qp?~H@6zib(GkS=o>ar_&#44U1t
zp5Cu5ByX7la`akGeyuEv>R3xZesvLpbdpc>aVlbKX`)&tbwx6t_@bvEUHA^h-M9_A
zPdx`8=z}EL`^<AX5*u58!Q>9+*zgs{P4*+OsptbtXN+w9e#UdvybL0&wyy3qjQOD^
z;eV1>;hCcmMlko7+eT#!zG{xKFK#fN;3ZfZY>5vff5GhIW6?fU6LU4K(B%FY3|-s;
zVf(~UXZBRQGv+_MBF}mV7W(+y(-d6_J7DneC=C2=gfBG5U{+Bh>zhb}gVQlkS60C~
zkHc`GMgjfKsbb0DkDw@P2;wEbVc*KH@U^@E#>9ki51dAj?Y3O-5P!fs)hfZ=BZ<7I
zSq!Q8OyPx_4QT88rveGTjRNtT(|D2C1;VV#P;Pr#8%@&s#qS<GOBW>fk!j>8YH@3%
zE{CI3v1K7ebzGsACq&(2FVft~o21rXO^JsJs55sf#dwsF+mwDda$Fwm-rt9N?SBE7
zC-QN<61r<C;Qk#-7{%7k!hc`jMExg-Do{Yl&uoUw<_|>tHb!fsJ5ZLWj@?FMarz?*
zto;4~nw1oocf<lSkDP{-O{19u#Syhf5oj#B13tm^@KF=6P5vOX_pqFK9%H6QbGUTA
zELL2924#M`AoPn7<_wAA*IYByWw|}Y840j=t2)*;u?%^w4la0_10vmuxUNwf&sFaN
zx9E%Tu4gpP2~gyZ^^HJTkqb~WJPp<~%5gagj^Llz1wJd@@#FiXxkP0Zx;H)+%-psK
z{F-hE3XYq?>hzi1c0m^DKRL=f`!LQ_Yb+gDyN~SJHxk46@g^%~5r2Lo?UUU|1A1vx
zCUc&;LZ0$Aaenl9u{u?iT&0NOl~i_tr{vvbbo0R>^6a<{jc^zAyjjn8u^zrX@fv1L
z=!Ak#Pr*%01gFoE!4c6tAiq}yC5ts#=39;NW-Rg7XA!*gcLa8<Wlqw;d*Crq4}GWn
zhk-4hL1ft|+;UC^uSY3jQ07$hnP!4B<sH$Oxsc|2i{hd-0Zw7;nB((5!$J1}_-A5@
z3x;e^$?Y<Dr0L)avuhCStcB~h+M#f1K70=90Ht|aDE}u5KF5v3bBYx(dv*X!Irai3
zkJ86{`9`?3F$-Qs$zXWzEr=Ly0^P<1T<sqz{LqyLUd=r)Q|vsqAg>y}#a-ma|JDY_
z|B9KzdOQp}5da-)Jm^h@F|Enhq+IDnp|f8ge`vKPIoGHN4^PV%m}Wl{J}){-Z{BSt
zPW?Vf82sbio7jC;Cx(8NFQ>uS4>U4pfV6DO>B_Wl(z^K&4y3$-j`h#s!Gi~Itw|HN
z9#mv3hauQJ)By(9-hxW69EJ}60}_nMntDwMOCGD^K}|`#;wC_uLnCpSg$?$#Gk#a2
z9rh=U!2}ywmUA{hwH<ai^T}lxc2xv3a&JP;fCJ8rb-*#|<Iw!N8z!(^$m*j5plfG~
zL7x_2A9E;K=m&w?=n=STl`?KT<b^*|bWqLq8%R?t1n!@PVVg(c+?X+_J>C&_vRt|U
zwE_6dda<t)b#U9ty)gad9jKl&6Vg{a;cC4CxD^i<!Xn!y@R(dJ@M^Gzox65HT-HwL
z&w9&?JxqqG-6ZUL6%FYxx;ek23;5FWIxypVK38=6KhAi`8EW*dBGs+{>Rt4aZ!^xJ
zPU9%@yYiDCbNV4gHZ+hc?WC*+SLp4~0gBF<N8bK}6s`RdB=+m#<YvZxiI&2#!A5Kr
zO%5f~hvAZFWxP@K6y6vbqg(a}OnEjMTepwE@4rsM5#y(zx{*NId>nI-nc&o)Q?Y01
zC>-Z@5^CK&Sf;WPUfx}bvD2B$RK|gEWtdNL&;egEPLg%@98|Ggic`@A_jImCU&mlf
zkPgMjx~aH--E{QMI1T$!FT;<=N1$wQ32HCU#Si88AnL9)Y)%bi+4s>{)b$4nY{z2F
zg?ad7&3MLTdJZ3^n}J@B4|IQKOgdJt2++=FzpIBE@mcV!UI(t8_yg)}9(s|S02E0S
z_WhCIgX~HnSUH?}x2~n9?yH+)_uc0{oG{_X9oD3|YNqtWu$|sp7*g2&w`ATRLz_O#
zqZ5}-P}PEF8rHj!EQUqVrztCF-i87aJCaN{7u};wM-%+x@E^u>so~^xqtLoaopFw2
z(A`lE9TOQt*v=9YHgmZ7#y?nIsE!AEEHG@6JkD1bji%M_;f~I5yyQL;U+n?3nFuI(
z%nA2j8;{N-7vWD$TO2ms4O4!(;)y>t7*fPIFzd9j$W9;o7#GziQh<-j)?k#}eC)XF
zj4pzW`2FyFJgeo2AB96;KHd_|)?1-(YaWdMd=jj3%izDaY))Nb9$I($VuD^3*oQ3O
z7SHd4N%GSm!)P?Oby+=lx&MV>?vi+RKpfxrPl0TW7U;D2!Zo?DOuPS8m|BtwUMFsV
zFgY8{^h7{dzaOkG<WSWkcWPQ8F37iSA&Va7ISzQvscm@6yS16n{a!^X*-+1)4eS>>
z%w5Pg23pdkvnS~4i;3hdv4@@o_0Y0!p_GyHl4{4uVqc#%o_Wol>mfPxlT^mdHX^w7
z?_YSbfUTWvW3c1wBy=Ayimw=#@qB;<_Ok+VoQnsJT+Te?qZnu7=X88+EQ!)NKDZ%*
zv6AF`aP`VrsH5(J6NZOlX~AOTk1~g(oeh3EAAzR}XJPvk6U@G-ha)7$qSP5hd>A|x
ze_AZT?7_3JHRLp;AL@iaU>sBC>^$^q8a`k%#NTde;ENr1KtfX$tyLL^XAkoTjQtK7
zyF-{0QVZ?(<-yc^b9fqZ$LfQA2?Pd&LRY#j*)a#3_YYYZO85i!{IvML2b|!m*D-F{
zuwveDw+{DqX%bn@Q6ih6bgp;oX3kwioSvT9U?o%SL<?%A8OJ+?dnskgJwDq+rYD#$
zKB=4^y|9<^Loe{s*7vBtmu2}S8KY)uGM$#Lrs%V5ZuzYNzP0X!)jc}c@K_Q3r`ThF
znJHe|G6xO%C*s3(Gq7N^A(p%}!$Wi4!>==T$iH?*buC{s08LcYScc9TW_awBI=1}q
z#%Gf5xD{t&q|R@sXG|~i7xQt?pIBTi#T=V#ZYWn*1&fXJFn0;#7tc;ax4@&2z<i#%
zYIYbXG{p}p-gxT5W<0;b2^G2~q0umHoIi_Ye=kbll~39j*XM@1R|nu`+(np>KNU<9
z0`b3Nj<`zeI%h6wi~em>Fm~$;b{~!vT$3IFQjgwnjod0uxBn=&Z^bz<np($;bT>fn
z_d_rvHk&^-&zo)*jNvz{tb*8YYw5s5))yGmBdw5AoW;jWTup}>?|A<SpPQOZr_NpB
z-}~v&OBM@}jbdJuC_kEbu9$RB$hRC@yObO)Kao@Fbmk%d4-cL|tXVo96%LEz!w2Hn
zdTczZTI%D1{d#zB_Ch=(=ZiZNWYC1Uij>QOaKEAz=2-LaX4EE(lbw#|9hno-mCdZg
zYoo<RS$uw+aVw@9;hc$aXn1Zq*0xwNFN`)mwogIF#j}`;QUg~@c;o$~c+CH0flCUP
zvMi|?N=%o;?x~6RB49giC=x@r$#YOvdIcUh{|WvwPOIjlXf!pPjB|hQ0bXYeuD|ZV
z@&=nw^viv)4Vi^5p*DEePzgVHi9nVh3^v${V`cR(xcPo8UYAk8iHvU{+b9idHg5p$
z?h<gCR!xrL)?E16@ifBj4&NYa!Hv||&ad@4O&!(GC~b2Hr3!56X+a|`AO4<ViY_;|
ztu>=l^*2biFPcQYcJdcT8j$pfUfSkzh<Z1(`jllCSS#6Mx1j+4yq}Fzo{hyFEta^#
z%L(h=xM6U@G{nQ^sAA-Xb8Y<bb&d@d{By***THBo#P;|&32bM~uM1utxVl)6v3Y(#
zIP<y+ZU4jZAEUAE&McHOG{k5A|G<9xG!zL+N6B@7Ix9s{B_sw19{-2eYZyyijl<i#
zIOC*g;XUItv}BCrHHN{czdjZ>__B<fO%hIEGeZ$E!*KK<n>D@_in~_J;L7lc_|0fK
zP8rOE*`$iIG^Ei#m@%%}_wbkh^>AzKR`bg~UWE$vX{a^wF6<qc$nr~75VcVqqGySr
z*UQu3bTtFK%~QCM8)b!p%>9tOLXXQYAwh$46dk@>O398ZD6vw9UX0D6A*C|DKjItR
zUUrY5G?$Dsr&8z04if)bNMj<hXxGjyG%ENbh_j56sLd!ex}}B79?VCpegW!i2*z)t
z?HF5|am7}6<6?%?sw2eu<JJtKKLSHP{Q#LojFJ3vA>Nglgg2Q_XV>C5R9`t8&(Dm(
z?JVavW<ny~9JIj;Z7bPpvBMwp*^K3P6Z8=<{)~$cuAKD;mg{k7Y4;w~iq>L+`FhlT
zn~Bn?yHRRyF#h!zh7AoBP&{V_X0;mQi1O`dFkc<<Y(MMcu179506#HD<*}XSjD06S
zSDy$ReC2>Wtkw&j#V{;?CJz70)_;|w-1(~&{JF12_|LqUJDWWkGg@k3!2l15GB03P
z?npRdss?vP#e%Mf82D`!<tFU3g244VIN1cA&l<18sXz){d2o_MHGh&^`8FCmTeRiv
zZGG|_YfIHNxpc81ku)D9)2AK5^oX5N%+|WGea0R=7*p)qkT)Jh6~^GT#<E!ExgR|V
zw>BWU|7G)xC7$TiqKd6HGWbv>0)J&P2Cvj_=DJqJFfIb;%~i$@ubnvN_Go<0e*U{A
z8Gr5Ej&;>@u-<Pk>U;}ArD3~J{lZG@xh{r7-}}MuV+!70vjy*p`(l20DsGc<Mbmd-
z7%+J+PI?)PW^6`I+j26Np3}tAa69bo+rsjIoAKQvXB;e_jB@pkXz4ZqJ-rwYeTM@$
z4;RB9Zy75@?I{d<qmRyt@44HI%`in}5>9lQNkgy$wwc<(dV@x|As9;&=ZjEQt_xf|
z+{u3&Vw{PVr_E<v8fY*$fMnaP>4@4y$X|Dmg5+;g*Q7|Ay7eHx|K~P}8Kv4Be^-)!
zp8SFAI)~^?a4wx$n?#45ddWO-1PV|2F!$nmlsK)4w~kN7j|Ij!+<g`9Q?|enCd+Y@
z+A0)mn~EK;T=9}*H<V9h^XP>^=u&EoI}S`jw?afMmZZLuz7A(3c%%2$bvP<~6<W>8
z#^dU-*mNlsB{)DM;Zi(gW6by+%xQWk9)ookVscarzG&Zqircf1n>zy)#q+Q_`U~`?
zxZu4RyKxR<FF7c(KJE7rxbcQEUi+hj%BJq9ILie3<3zDfEfxb$JEF8872+0OfYmaM
zAd)HsO>-N3*3$tuU;1EuxC|C$8)Cx7UTDZ<>-DE>?(1$*%vhGdSxWzevjxX3RaUjZ
zw+KD5D>kB=Cc6bSaiM(ck2)?tQI2uvK61L1eO7hqkA!-z_sQW_Am6H*O5s}dq<`xg
zl}`1h^YgyY%cUR4;Da{>9$1148Ao8v*R6Qr^aRWs#aJ6EMmTcU0(92$Ko7%Egvn!3
z_pB0rmG{S54|i-C>4F(A7GmN}d%U~C9-mCKML8!A98oeAyO~EjeA5c_tDcK0XSU(v
zx>?K@u@42Fx8QmDa(orI3hRH&N5$A>D6vBuD?Fr7?^O<ZvrOy5=s4_{G#p=*=VI^7
z6g*PS>e|(;F1sKS@jxuD5%s{ct~+qL`F`+^AC7UKM<ZVpjw#05aP(Cp)c)KFU(Q71
zs&zj>deS_UsBpo9RdU!kHV($!vBUY9H(=oT6u!U7g}kO}Vt`c@*fGbvcgze@Z%qS_
z*i!1<ClCK7RFJ5|bZ9%ckN-1Hjk9`yp!Fh~kLsSscW=2W5WQ$dBE7d8$2l31yJd$^
z%4iAw8BxirUJ0R%j}=>HEMv^p?;}w+(G&+q>EKnyGmuKq$6M@7T}#P$GdKp93KQ{w
zM-sj(h{b(2b1{8-0G2c6dACdoT1q(JzDf(6u+A2<`ol22Bocp(<?v+OJp6HJHg@71
ztdtAHC9ewc?~~P-(dvb5dS>{JKZwtY=i`%(e6-(|gE`8<IHF||_8Y9lE;So;^-9B9
z#-(wqD8Mg@pJ4JB=41LB2)Cnu!8n<9XfElDm&UF^edT0)xSeGW<Lp5H4$G>j&cqIx
z4B!&7!Ai{<^yV1SXO`1{VO9%K>3iXo*dLhmzLTpN+7EMHUxUjgg}mW+Wxgq<TX?Rm
z9-@97hdEhcux-mRDynyevFUSZ;)q(VEoc_4Iqyv)ip1$rlrbr}=JWCM{rLM>LcU8h
zDAOR5o`<y2d@FZ)_jD58>9<Asd75a(O~a9MebC5bGJ5rzV(kbE{2?8Nyh9o~osGtE
z%*ivsZaFU9v>Sh&vc}VBh`-LV{o@y7Bxq-#$k8-xGT(;f_MTXDy%1N+TB2Ys<4w86
z;qZUV!+BvhsulQPT|hEEem(-9xGJN?qxslV6^~yn*&QFV5ZdEWWbSG_Za)MeohxwT
zNefh15r)@CdZR$v91|az;_rI4Unx(<@1CCcJ*Ee?ED6Nbru7gXasW!iTcDcVk>5st
z<oX)|am(hp$TMc<#(G)o+HFI(d)=YEWEh@a$XwP&XBdxP2^<88LY>A8Qc9{2Sn7|Z
zG?Rm1-<JvROtz7){SGqDKT82$?$D~FUle(JKVPLWmlk9%qZ;8&`Xvx;$+|m{)CLYy
z-N0^Y)4M`<cKYB02OU&pS=y^10r+fw7@nNF97C3SVP4O0oIG^`>ex&{(Wh}}9TkP2
zD|FClhZS!7!dS?|3sE^e9xE45#zQ%VXylxa!(th4lAU)ij&?&S&qOQ~F2+peLMm=`
zM(an@&?9aw&i%X)E#kw_aCJ6*ypx7?N_wc-v>L^(=VGmF87fT3#TE0X;ImX;{5;18
zvoz<TR-XV<>R84#&j9spib3#=?Ne`6@j_HBaG-}jGt5!p{SBBrPX+UCECibc?A|%D
z1&mS?AX@VS1hLvL$L2dIM1)h>YfY{tQWzHUn(wI05<q+th&&y%(p)D8`yC#`)~B^x
z{kUAXzc7~E_oY(jYj$^fQA38^hMaG551Cec6WVTY7x+fpAlI?yNJ;7w|DJvKV<srH
z_%DA+i+m<v#LzUXxibpOL@jZfqZeaEdElOh)A0V6Wb|U3w1PBSj4B<D|I#L4{elU&
z$a4{@zLr9BcNZMcu))#BE3qgh0)2N2a6WStHEHT#%qdm$d|!+)D~s`QV*-8_B%yP%
z8$O!$hGp-f+1Y3&Rww!4_O58GG^#|;H@295EECr=uJ!Ecaj0mr5+i!O(6ne5u9q5%
zMT-Yv7<1Y=Xo(^}Cyg;|#PL+<bWD4pfs1zPfRw>^n6XELu@gL?D}iyV9M$o>paA9!
zEQGM%!|?ALmY3BYOUtT<An3|eh<dXT>i+!W%x3?`JWjts&``<lwC&BdS;GV)H$+iE
zZ8BeP9!}3!c2L1dQ4oDtLm@AY2}XU*rnR|+WTTWt0}*M|6)e_L6)oEGWk9k;k!5){
zWaN<6aetg)>V;qY=3{?C2rd%DVUbEU{>}`*2aL&adS5uUt<A+TW?{JG;x05Ey#%W(
zZP;^*!bMK2u~~XO?kHecG>0|#VR;U2VZKAf#Y?caG#J;J=3>xx##FnSgg<Z2#`}_s
z@n~Kwu6;8P_p}D0&8`YOdVCi?N_NJ6<}dP*Vl@EWT^QM}g$qW;VXD*y4CFVW)B70Y
z-)Et}Hmeo*Tca*>z+H56!VfprK-?*IKR68dX1)ow1olGvDi@Z)wn6idHgMP{iQ}%k
z<a|7)W9ox5{H3QexzM;WP_&!^8Ao~OG&6xYb6#*}Bbxa2$K^03<pIcjIR+7FH(+FU
zCnq@A%^OIrCnd3${0~tj>b*XKBu0<tKe;R>5=-P8?%$_heIgWmR<z~LbXSUXzd?$N
zdnk#WO`8u)$MFvXapIRud~fcBOE=EN1{)8YJrK({&kOOebO6pfk&M^ZN1~^0G&-G(
zzyQXaX)jre{y$e@O0Yj;M(bk-KON`Xo`Xw*H>2mdxma{B9|u}eP~%22j!V+Ui`Pxi
zaMU>bes><0{uhVJmu8_1%dc%XuglyIJ230t7L2-Cf^sh(!r#*@lX<=rZypT8yi`|I
z{kIrZ_zAf9MFwu{{l=|)62sU*ak$);IWJk?Ci%%P&R;1I&9zhErgNoG)?gdDy_mvz
zNwUn*YA2kN#yAA5H;{U!4ph=JVQsbyxLr^JL)$_MpLv-Nbk!0lFy@?;lq972NAX$O
z)xxZ2Q~6rkYQ7-q1SOAn!!>p9AuT5h2uv~HZ*Cf*m{b)(N7!f5i;|}S*Jb2tE#x1*
zXs694$vBVg(b+m7xRW_Y%O#Rg(mDanM}?xwN%qXfZ^Bab0}R-lfms{3;k2b|aTd#k
z4;cob+~Qc&<7095iwgX_t_tsK39#d*8hU-$fVCmRQSYBRE*B|hS;>W1FMkAW7dhb4
z_2Ky2Z53lmF^;^UG;&c^__rh(KRTsjP^b#tyA_8Dd(H90O9k{zQ9warChjQWp(<i8
zPSJc0QV|~@CC?7){)%BwUJE?9;ed^W@z|1f0vgf|!mi$9;4y)@LbS{{;SnE{?c#91
zSt(pH`^4tlX5y_gW6&=ixx}-Nz(7_4r^{=D`ltt>_RWExerPpSSiRul-|i-#(>G!3
ztQ)j-X)%?IbE3OJ=U8@U5q~z%%<6^qN)nk6)8sz=nII}Jo6h?7Hp_frz13=o7E5WO
zJmz_BFmz*1quD6cIUhyx67Y3yG)4?0psq<O77pKxBZ5*G^EeU}b}+tv!F-%5xdzW$
zvD`n)={KBp#>tiAQ1gBjKA4z<dv4g^)#w21vt7tCKnwA0)pR`fVHA3Yeuu?-60!d8
zD;W6CACIkbM2S#V+sfaG9n+lAJ}n*hE~>!MPOg}id>98rQ?cudKZZ_ffxwxb*gK;b
z0{+cIed+!1_ILr-I~~Bx-LB{|c`bf!H^RKCI9ybtfj1Yc;-{+bP$s$#5BnX0w@WSf
zOQK`A>yz0&>?w&|m#uMdM?csd8U@Mw${~HuC|r9po7@%B1rnM@RA+pID~lZoU$cJk
zKXm^HU-lN$*$-m`nVP!vDP;rA-@2Pb8rP7?jO)}kYKSVPzNGVivPq8OXwjHtdU!mX
z_7B^DtFp)8J00fQde3wL9^rVnHw_hEX0UI3Hd>8i&KAZk%2~P>^JqUVe_x8x%!~MD
zoGwPK@xq~;I5dz>z$yP`;*n~`J&(^tM~h*2>B%w-yJvxGUWenKU2{;wb0O~7v=?v1
zx#H$G0{plo9_1fS!i&}apifj5)t^3rm8wdp<0^x)<BjmB+d;glw*Xxf70{*C7USM(
z<AOcLD7=}Cp0nd|(0UCzhLpm|-F}!K#ay$x%9xw@55j^XFwo8v{cj0yTk#C+*>A_(
z3NPW*^<r3XKMyuE2SSy)0Ui!t0e`dfF!<R#zG?MG&{b4K)6gi&t7hEA>zd%m`lfG+
z55QZ+Mo#m%GNsAO5#yhc{*hw(G4m4tr@eywr2ToHxed&P@}8unmeFX%^Ay!4Ls#yv
z#f}~y>^c{TB5IB}?v)MR=v;&TgLCjuTnfr)$6?Ql0?hD_K*7dnEN=9~XZA}lC~`7N
zR%_z#xJ>+gAqJn^V}0jATdZ<kg`Ey_u`9_NxpO{v=9W6f)vMy(#JO0dmW%%tI^*m?
z6|~;D4!`=w<I;W?+%?n#V{gjgwp01&>op1Ai!x@?<UoAv7J`obORxiXfu8L=)HL6K
zzCJtg!oPo<X;MG5d@{o5I1Nnqo`!`E;`qX~9D*7nK<k7C1TBb#Mg2D5v3d}K|G0sD
z(Ic38D~o?tJpqEUU&14!Ww2MUmpkTV3R(8oz&mz0+BH4mPVK%xKe9Wyv~y;h;*AVm
zt^6gy@w?pF7G+xHE83zi7~WDkB%r#a<Ge_p0Uz3=M8%%7X}|Xj8n?599)~`s2GL-2
z_~d|bA+DHR9EGD_mY`p-E6O+-;we_Ej#l)+fN9xi^d|<7wyC0GupiDiXMyFcF6#7t
z8Lr&14l{T7;F{H;*qfM)a-YobLR~tZc{BwtyF1{$kZ63v`1<bgs;K*%2gwSS1(x1{
z8%L?5tm*+YbdA7|^4WMgrWEs@hv7x*Ae3cyvz5KZXy+A#M^jbMVb2V_=JOxqU3vn7
zb#w9hyXB}}6@_=FWWu%4$><TX6PB4SM8DVWxaFNKj*$&O(YqC#!3O5gE?x&m%I(nP
zod+zi-UIi(%3$jgb)0!i100{+f*pJ2K{mT1eVMO;4fD<E)V5k~Y_}owpFIY?_Eli`
z-x2P0@@}a2spJ|Q5u#Tlk*}*NlpN&+V_1&sbEzFw-s_>dH)7;?og??Zt)%1Si7RS2
z^w<)LfA=xwutqY9F!qi{D)Z&NUxX(Z+ao1k7yrG>#94;Lc>anvDt<0TRhg}*k}5#$
z!_4_!z8;mQSz(uG1}6EJ<9CNl>~br`-wzAX&&d(n-t|H0f|<B!`er;`ZH8~*Jt#3I
z&@s*VIC1<OTpPU>)vQG^`$Q2Avl@aCDHd4O$)4Sf_aL*Y5aS2_!@taTJbsS|wjN!C
z{uli5KYzxrjZ{W^r<J(+^GGz_E`+M>T5!E91xB|7p>Xy%xY%+QW{F*ZMMK}A;N2*c
z-B1DU-x7q+4bDO>O~Zs@PtsjrKt*O{Fz0eQn3bL6-6lj(VU9KocjhU=>MozIJB_Bk
z(&he)RHr@O_bJkJ9Dm$dl4>ItL3n#D#rSt|`N`Ay_`$1uow7bny;eofCYoS9^I7lE
zbY^^$5S&|QiyMC}M3><K_||9>-e%riwT&#7%ctU`hCFQOV>Mz;0Z#kA2qT8?MOEEB
z=>K{Z#@Hud#OxTnI>Z=shT6D(?>-D-x%AiVQD}YeHrSY!F)kAO_9vO(!7-gsIa(2a
z-0ueNng#xf-HN@#TA^D`4l6a<!J}v^c8?i{Y0b<Rwm=IV)LUT0_5z$?Tn^6{zlR$A
z*RbV?2J=^|<D_R^*rj;~l#eDuXn`lr^<9n0EYmx&$qi(0XTqE<D?x2+DVH@g2+nqn
zsC6oWo1Vmxu&e~!m00gj-GQ^M?Gt!=Z35@I0b!^7Y^b_b4FP$3`J{XH6r^;Tn-I1f
zycL?^i`aImIX{7`U#I{B3#JNZ+^DCJ$HrV1n`wPA=_~cM>hV>|A@usB8~!+>juAFK
zctg_@H;y*I8J$~E!J!=Au`K`Io&eNP3db+KvoKN`vAQ7?M{qmw-Gn6k)XsS3^)a~m
z(?N8M%)s=iVJLq{5tF)BVYg`req0obbJ^dBWzNEaZHf5s#7I1n5{4FIw&I}I7x?=-
z3!8kEalMHx{=8g?Z~s=HFmf(B*oR@wCw2V1PXSj8=V3-U>p$wRM%TCdaCy>36k+??
z&I`k^k+JCQ59;8zExF*z>I}vcN8-`OK?sx3z*DL+pi-lbS%azI98(6%W?Ny=*m_>r
z_KEMV-^t6jM)GTS6hh#cF90D5RFEY`Q)KmtTE4*YxIQ>lW)C4%-TXbt{UmbXP1D4@
zG%$7ez@6Qk0~5D~&<~uzXX&Z%@h9Jsce-{<+wt36*=S1|loM;&{bVhsx^2T#&m8cg
z;%<}=S&DCTremN?AUe9OL@kR0SS~gi*L~fD+v3^zB-0az(!DWxRuU$yo`=<4%+b@n
z1BW74p(f)|@Y_Ri;pe%ipq-14a}j4RWwR9ezd-Nwev~)<331{|7!rOCR1TG4K@s9*
z%{_SVlPAjOc;n>j3250l9T!C0fV)K#(Q*5J+>u%jku%(pTVjaAPMTt-P8f5`_koA=
zczoF<g4ShQ*&MMP#!6Shn}KQwKO%)a;&ZW_^&ItMm|JS)7g+nZ9Rya6U|TXr5Y%-U
zc6~Yl#kFh=QaKO1uRh{j?i9d)vo!>{9)QI6G4SJuJ+)3qgt{~L_%U(2pnc3i`1Bx?
z{I<(-ooXjZ<I+CL&l?9_Qdj813e}d!S6R)^coY3wJ-p@VdG(egjWyVQ#fJ6Zf-#-#
z4TX}%C>qFNPUcGd^2`p`ThBuOstGu$dKylunTKIkp2*jG;N;j9*jCEnr=fVpJ+MWE
zdKc8JT7pK40~otxKK?rDihGu=Lq0Q%u@3#Pua?b8E&B*-Crv})U3<*9x)Hw`mqG)6
zfvc-haG{PWc0NzQ&81s0Ohg8^{~Up-rz>GXU=Z$R??WrWka_y@@yl{8Oi&$*D@%8y
z(%m@p=;N?)fXx@{wQ!5ghGTJnBG?478Mo~dvFphVKG}6CmU&!+r)f`M-Nj~j%JTWP
z4*kuq;`0fYC}U!N5x*eP4jdkPlWE)u+FsVog|&_lL?#vsbVIWQH$=yi$ZHR)b{%CJ
z#<D|CFTH9miRfx}|M5x?UZqWsZ;6pwjR@&(?xa5%MdZ+-PQv_!*!?yPwI)R3PsZl0
zz8Qe+?EJLEBNq)N=A!H075FYN6us|z<Mjs#IPU5S+%Urm%}z|k#U~1I)6Ga+JTnis
znvX{xedb45dYE}iw&0B6?4D&VgPQ_3;I_#daQhs8bUgPNio%#H*&z>Ce!CBy>>UuR
z0dyo?1Bb86(KovYhTU6;Pv<^`^GzbCaP}{o%Z`QGjlQ_XfCrI>?|^L0ng3u18jp2C
ze|s}@m9oO&Z~uV!qj+ozQ^cm9c_8g}9CmG*k6R@SP+eRfpEmXK8{eAJ+nQ|ny6Z0(
z+lXPe$`Oe9d%vlp?==|9E#P0?@a2!)Z3D$0*Fg|J3`HVTEq&*F<+j>q)2z3J6k|CJ
ze1@A)eS{%x&9|V!joT^4dJ88JzmmFMC(`4137Bmy-*R2gh|K<r!T!%X@NIiEHsEHo
z)Cs_jLq>Rpd3h6qL-42La^y$(;%dJ<<_}(jA$B!*F~|b@<ntIWM}Y0J!!TiGBC3y@
zjeDJzqC4Zt#~d4kWv^mzQgkflpHM)5ZYkC;_rm+b6!76K=2Tcxhz3`za9%x|6JUE`
z+2!N-gYmils{Dl)$9!<1&s@~}y&nXnvKV|)14pUk;Hy^VQvV(eQ47~#)cbDGC>V)J
zcTF)PLK9y;GUQBFo`k_^A7Jm8MC@eq*%Cc@oTif>27VzfYOO5Rswu*r>N;?HQw3*(
z7B}nQQmD48VElhm_}W$QzsAl4tfu#G_aRAzBvg{hRHRZgt?x~elt|J<Nu`oXlR_c8
zc|g)Yl7tXKX6^Mz=FA~;$dsu_KSQRR$9doLo_F}4>pIJ|+556vYxS(P_I}p)d*7d1
zOfWo3CkAb$jL}-+u!fC7h58C{dGTthn`UHnYUfy5eY~4saBdK@o>5g5z4#V8J9;f^
zAnu_&+Lu*I97MT0y~(|;ki9sN!el=#gr<b|j#p2^(7r}Ep*a=ba*R=^N4Ql#nfK}^
z;Bd(zw3_<B{4Ur1H4H&P`U0LUH62P%!Vna-3b!4yVG%bTCSrf|%CiRP`JiRQMC{`{
za*dtNIJ&_XdFS=9v?UFmTADb^&$WqdxmbR87=|950P8&&=(nwvmPTqLW^^Cq{>emi
zr%l)<8-;O?cy71X4xz(*ImGRiRIqX|Vkenkz%OqiexKRr<527md@LmL{{1(}K5BCq
zkA6xw>1y9?)E1|J^6K5RLED&g{q@;EuE9QHbc`;n7Vz|A4-_t{WOZv(1;t%Qm}bpP
z`cd1J8t+um82L)!UMF{U;vZ|lTIC8ES{!CmNtvy`>r3t{EQpQREl7GsNmZt(lUjDU
zV0^5Y*-!t?B=hyzahC$-uyF?Ne~87BDJd9#F8~Uf5-2qq!+DTB+)p^db7KG=$F9Ta
zc>z#*Fa`T!Ijj3nCh7;2And&bJPT6sv^)^ErzT>IsTB&eC*bJBKG-Nf9rIdvZ)kQR
zW(6+8?$*gTAGsOL{QQ!+Iu&O5iBMdBo38zeLVT?b^skm-*u;K#>2;94yMCZISrSZa
zipP;piRkTfn{3;T&?wFRyni20as1kMbx)@STuW^4mPE%ijUirGKut*^srY=P56vMM
zc={4O)Ga67$>kh3^MbNxr;>eF?vp!rO|U-unaov{nJ)J|mOXewwlj@cK=OG}*I<pP
zedC=tbInvb)%uHCFAtYKJdsTw*Q&G9@I#hokLfX|HzllKXZMruj;k{b+iR?G@eeDz
z>&|4nH-NdGaAT#rY=tKKG|YEg3-)>%`oFM*mQ@PA^;rV>_r7p=;)+|-+%Y(64Cc(?
z8I2XZAL_IdZ{^2BO-Tp6jW(e4;}q2V%E7D$hoD-biK6F(-o68&>%R>Srs3!@xg3L&
z=OJXrc^cno4K~+3CJT8Bcx|19GX7aGi?e!sGs0o5=7HCTw2^A)fXFa6ypHmLePIPv
zMlHtCXO2+S8v>=NkwU?MJ*1kn5raF0VdR2Zn$qt9J(*;LoGzABz5Ro5G)srHhVi`e
z-#t;dJxN+-Fi$jAe=U7;-W+{RdH&6XB>EtChq`>yL+!JD<U5lIp0zuu{-=}BmvfNn
zKbwj2nLn(q^X$^H*>lCXB3ITP0l_&|LUn4_NYfxotm}8o>ip9xR@uELbDGqbe0M(*
zRG-`uq#om$&ow93Z{121-JgazGHY<TQzVA+%-an)6R@Ed&)cf?#M>ZK6n5l7Gcpld
zmCHECFA7#61^E2V7|s1{5ucTbVA&+-AD@V&r4i6+;CYk#CPVf55F{;Ff>-yNNcY$n
z<eare>idz<iJpwKJ}G$WwVb}&s=)Wna-Kc$j^>{#gFk1dpEsF+=UbiewV7wg<o}@!
zH@0$S-zsw6$M5gmjBx$%5z1n7asFp#+W54QbbLkn#&b>+vgHx5j<eC{bjIhmGRX4F
z5Vo6zk@81x7!5x}La;uXBlE@9qwBao_91<sO=9s(J94S577Pz4Gp7v+Os#bsNpCEn
z@ZiT}ys(l@IGe%zx-Vn)%jbxN5vgp%dr_2MH<P(v$Yt7Be95B9lEyf!7V11I#Ngol
zVsPX|aoG16Oyf};!Y7QzZ4Me)8lQ-&GIQh!t{8O77PcR}kvujVJ3jNAk?g5>8XAUG
zmBVrIw*;HEu0h1FFpU3Wil~NMeC;_7HwSD+)usv1nh*fvG*g7;563$;0z){rFGFcC
zlsWfL->nk{9G-$=?**`J8ixC$^>~JBE9KR5zNKw07}pf8>5@-{>;q-Eb%k-3F<c(`
zkjlGgNc-)i)>A=P|2_`OUIJ?#slktH(rjNi)1kTy`cTMy!Hv%a`3*y{C%~QON!myU
zo#~H_Q+gw^MZxOPrd?F1c#vF6rR=;}Fj@UMK++ZIw9RWb6&v>#G<)r!#)mpWr<0TD
z{_x8r)Ah@VrLVt9wYymhD}LFt;yioKoKoX+tVPo17e%bbK%42Qj1$ku^kA9Zne21S
z4CcG&D6^Vnip-fdd~P3t4O*FqUAP>I3G(>9*A0V}SEGi{F^yFCcmAsoRId#}DX&ZK
zdwT#cbJszoJq3rk_VMGbVR$Q}h;!})_mI(?1*?wFQf-{G%YnI9JWBf-qUY{pELgn(
zF3-B5pKlUuey>6Y4aU*tcl=y>M|0&k2eot&nvct2XGkFyoj*%bodP_#-xW(Y)YFBN
z7wG=#-Q+q*6CD?tDCRNu3+W``wC-FuG%I4;y5&f(P{3uQShy-(rES?d^rC1#$@M=i
ze7<^>d^B3fr}rS1@GX_<ChQdla`vX)Ix}Hsb)r}(r$GJgRf#f9wdRdiQmTx*q)3aO
zM6ieT*<5pW?S$)91<}vAoT}dCk>SEDHYCN2y*ac+v{2w4-qwBWv{=aGSNsrS>(?{?
zg&*1W2N@`vVg;?5By1lRgd3YAc<U8|{Joj5`5J+j<2VDF*L&ahnGUPB)6j239ApBf
z@p<byL`AJfRC6LsHcB9uWy-mo>B!XOV?1v#=W`pP@%}_OJxWI}x4#*9d@L$?Ke<mZ
zmR{i3d{!9pHpxL%*oPE5Z?qQ(NGtkC&+_CkMm`Fz0qgl(_&d4w8i>0aUeSll`&2$V
z6It^*gzLUz@L_i&{jSkLzb!^w-~E7;xQ5>AfCv4Pbe6p5XV8x|r8G`bEfg!O3DsLF
z#JhnnL}iP)()J@QLVkp~(EjHzXEt;aOSH3Tw~7+(`0b|l<G01OVL9}0s2p_l29kaD
zX)EQNPvWf1Q}ic(7|TdqA(X~nW?X(mC+2k*lS~Uq`NwZz(!&#Mo9%e^aaAR=YS_o1
zTZET4-Eb;C9`}00!7MKdcO!C;yq1q?=yF_7cE&NMAoOkCf?X5xF|Y3wxP--{i&+Xp
zwPJKW?t!rM5)9$}<~>(tqt-$eb}n|9wBHLmIa}6q#B`+nvlR21K2qBgFEni{httEc
zsNX#oo^yB~`ic>*sVqe8ibyoJY2f~?wQ%1u4@$bRy!MrWj;OxqmCCd1n`IDTy%e8!
zu0hmtH&k-odh>@AQb{-@><s5D&)qz$hxb%!oL5-QzS0Z1E7#DFl6GN6f*Lj5)FB`D
z+v2qJ>l9MFOE7CaEm-_+x01Yd7d#_Mh0EOMbXCqnY<`f#_GbU23w?9M#GJ{jN!gdl
zgydLsY|mqV{^XFgvLO9@@v2Z$<S*XeH%FjUccxUjTZ|9)VzW1$XNI<FRBU^kb*}W_
zEV9K2s@Z`fU+oc@zZ$WE2h6>LaeQtlmY8hA+k4A#cvv9PA0L9b%TSCynhon0>k!i4
z43A3YK&*6uPuCcP7Ys-LzKU3G+8<ZU?eMF2FT`G$3!OqsXnJ+U={^21-;sjp5d~P^
zUk8D{Jz;;QH)mF6BVorhWPGfkA5FbbsW%6o4_RTzQb!E-cE_W;8)3`mWnCBgW0-;h
z{1?r_SgyZR_)#kC=UVhq@BPBm5Bb8rFA6Z+Gz<sYR@1s{4LsyIX^A(+;*Cp^cx>Ss
zay{TKs0F>XdRFn8K3$zkUw<B=poCZA{=wXT&v~3jG=pg8omH$P@3$D=vma->bZ4da
zc355HtU4vz<s_rN`Got>ET(x_!qkR77F?xY+0=6(oDmpA^Ul6z>~4)%I(mb2MROAL
z2J`Px125>SXJgjsAk2?U$HWv~W7g)rkUl)8{8<e24g{cHXFSX+e4x=D4D&vm_o__r
zdEturA-mCY;X>T=?FWy#a5&E#hbfAdxY~Ila^qu>WATwzj!J{$R-SEIwg4wHdH!+n
zS<+qeg~}^SXi!xmb{mCL)FX2w^19e`Up;(wen3snGmw3^gNo(5W9Kbj3_2c3FP^v2
zjNQh#-|!DDIT=7b;)C&U@Kdr_-50mLO*uR0C+%rb$M<`-6gkpUZ2y`p`UJNKE{_lM
zXrF7OtT;lfy*`ZOcKo0LMT}<4zY;60^~A-(ePOA`Tj^@sW-)h#g#G&3pVZnri$e}y
zC8x;6Y}Bcvr1m0~6uyt6xtDja**4rg-gP}I(pyOe-8PDg-^elTQ<2gxwmVp@Vg{b=
zUJmnA2l%*eLHb?~uE9&kjr%r8Z7V@dek5*LOOSZn1R<Z?AXzyNy-Mv+TBQz+U4-4m
zSx`^cg=##14s+<Va-T)5E<EGHFnYKTKGddT*9Uv7^Y(%EW$ta$(Z(^$AUxpOh2#P4
z^hJaFdmK3b_sCDWnKKz%Dh2Gf?}X)vdT5Yy!N|VdaerGbv^MmB+JY0LGBgqHen!yR
zIS#i}gSnRICapJU5iHcVTWz@Uj)ITh7iu?;rsw^<>HOs5^yKb-(eTJZ=6>*xSQ{jR
z<w0lYk-oq=4H9-~s2&~5j%EG28nH<A_d;2f6aCHgB=fJLb5))&{hR?C^7a8|;yH*m
znj@Lk8gG)=d074sm6>L*WcJeEinJ3hFwLR!NW!)MYV<>FUYR4*hRnj6US6o_SO}-=
zjkxGufa2H`yy4nK1??nwoX$m`!GZYYxD_u0xE3ln3l8lia9@~;-z^jHawyLua_EIO
z`(n^1i|YZS&9TtZ7?pE_xRz%i|9<;Jg|nUU(y)Sp%ervh>=KMKRY$W#8&`J4<Ib8q
z#PD3d*V&x8cx?^I*81YhZc7;T9RjvM4fnU_l8Ih2!Y0W>MZ*r(a}HQ(Z@FyQS9Kh$
zucgw}%*)i5XD)`H6u1U1NZjK(jfR)(qPf$`>FX2)SWnqVH%&&N`e%tSs)6T2G<_Fj
zyS@;g<|?vZm%0i&9WRQfwdL8vpOYzOQnjE|)+%JZ&LY+>PgAeF6-%~TQ-kLRQN}xh
z$zQ%9T<B^<m%T07(jNLkeqsjmuZ>}@E9%A78qus(2ouM=Tf#Q1*a+I?g^RZmv1LjE
zCcazDv$mGw*#HU7cu&W&G5mXCW)cjdy^;Hd*HStA)busioH%;JXUH^|=traCLk`a?
zEXO6Yk(k^~6Z>~hM@lfyfZV||QN!n8uXlgk?0=RLo#XIw(iZ9;07z9fP=n+%?QvO0
zvXe$5<(V9O_my+a<2_P2+8NgM_SC#w9>K4-ke*@<)kmBm>6w$9JsU{7F8-kSe&^}_
zwzI<Z3yn0C0x6&O4&;BE3HQCrsPtB?_~KSAZAeswVdt%s&ugeXkDjE`>UOg8e?_Th
zWazfy0_v%CkTh>+3npEcG4F>P*~`Y`;^H~OsMYw6v?527yKZ~4t$Qm-HT;pFeetc;
z%cxJnn-w281M;OP6A>oZ8r>Exd?cJjEz8v2RfzBW1y-?0nNFU}hI6+h#D!<U=6Wo8
zNAmpg1FqOL%@?v<Pq9G98%}pp5OO6EoupY96g2@qWDD?xYa09LtUz5n*D^exhwe{Z
zv9Kl=&db)KWO4)!hIoT18)8m^HpbNdq)&JF95P`Vp6rXlX)E5N?$Q|zu8R@BR0Az~
ziC_^tXVN+eCNrlad%G6C9^OVmTON)ck)vY6P28(FntMmPVr_#FDY^&HZZiqpm@ynn
zeLAQ!_jZgK)JXN-jdWkVH@%5YqS7e`NY=NjpjKZj^t7uci&>)unJ(Xi)~kDk;MPbc
zH9E*DeB#7L&52@oMhaVdbPrW5{x14zHw$`qwWvNSRcN}ho?J@~2|fmwNZG+c?C(@S
zIc<vUWtJk-eqJF;mNZ$)7Pzzh_sdz%K|8T>qm?+-;Gp<=T`^vWMf^T{0rqxG#m27=
z2p*n-M=b@2>Ro_8yzaCqWfWZZrywdW0@iiWNXZBQYYD>L*gz<9PiRY_IaFT$p;gZ^
zkm|3GNb^zX{NVtaHZFxx!*De2;M&Q&SqMwffcKTpRQk#oS;y~D;*)MXk2i;WmfxX}
z9|N)L%Lx2NFbNn2kGHXi@w!A`gEi@bLU(LdI!5WwKXWF0rjTJJlK-r(ur*AjuX~zk
z?f$bAaGmoK%q&Q9Gm*AW8%gHr);#|sm(*VK_Ze%F!7Oc|xn2)-CzEKZZy({!r1yfl
z@md--#GLu8>A}3qTSz-V5Q=)OqWksB#mj>fD7!41sN)#d{&9VuezKKsZld_hf4z9_
zw;QPi3}qQ?u-IUCPCW6`NJt&{h>hT$)X;T_?9$B;o@KrQkB=vyV<gYnd{oSRo*}&N
z9>8mDn~}3=33h&32Td72-gisJ?D8;7Uz7uzJubXP<p6g%p3SW|7I(5W@uQnMS{HJ5
z_Lg;+HNB4BDcHi2=5SVQBHmro!=WDw@U6uMN5WrGtvS~#%yon3_I6r7^AKs2nL&Nt
zbVN=HL)*+Ws+f3;J{ye37ymU#{le?_Yh-bf`=Qj{H_^_eL$Evf8%^`#+KkkHNLf2k
zXwS_f<COQDL7PM^$-^j3P=(8~lY->zf~t=3=O}pjOQG;wm9$};27A!+4$0oK5G}&>
z#SFs~>Aa++Y<x&(Ch7NDNdIETt_@z#R)^+@lKkybNy1sH_<~G!eX<@~d4DT~UGt{*
z@s->ItVPup%FJ5-qSQIujeWk)Gt=gNIN`0!uZx*mq3O5;FK5`|!~7+jv9TGaIi+87
zwH?CWgrK2M66~yUv9ihuXY2f6FkD2-c7H50+<|r>5>jyn&JS_GTvx77j$VUzGj?ON
zdj-bd?ukf`32>6Nhs<CVXrEsKyG8t(cbkJy-O>1(G6n`|s<^kn05c>!NB4#$fA2J0
zD!)R{{kG8V4f!~I_b@G1=bqA0ooUGCr}QhchQ6ORMVI#;G~&B9x%CdBt^By1JAarn
zTK(u2*Xuv@*hkI_V#Le3tEpR$AEdc0nI<?4p^VOEV(FllmidkP;;%<eV&h>osyW@y
z()qEOFs|2iacMfAuMb<tB2}6>Yf+muRH`!Tpz%!0(ttESrwJ8)ob5AdDf64KfpYo|
zV7~tU2<N2tg!(a3@$Sd|)bCsXQ+O1@I{ukX&l96@BF+!f=0)Jm>RG5;C4phk3g}yJ
z#C7?2$QVt<6|Qfrmu}&l_zawW$RNaL!hVDo*Q84jnX?R$q0w+YI0b$D<k2Q;gp-z$
z*eTUOb+tUirxQ@lePAtuI%W+vf&JY(q`J6+Vtjhw^$|y`^3BJYiPK@8evcli8epQr
zR`}?*(`YTu35fH?FI^Yx%-t=FRb*t|HJ>hSJwffO_tG9m&h;93k|f!>f<e?yvMt#|
z%D+!glVdpfXM7-+FIhs3{@kj{pGSqx+M8(f5nYy$bVQspMnd)NeVNQydAgfpNS7nx
z$=md(DF5_`=y^k4jGMcW+8t9w#Va>SvyUS?p{L6Jd|XBQpP4bG;MJl*`XW|$;Jj33
ztr-h<aADOlq0D1)21%0cS++e$V>P<V#PaPkQFFr;XRd~#WnKvW44IAT9<$Kv_$sVJ
zF>LcU;*(!KHu3jRRUT}OEYXk`gsrK@uy0t1l5{mx9OONgTxaaETZI+8$8@Yw6;Cef
zppC}jE6?Lnh>?Rsm@N8jF2;%EMhdU&gM>?32%fKrLfKr*T-T3hcby|`?r;CunR|k5
zI;f%LEUlFo!A$y#N^WP;t*jE767!HUz7^A-f;V)te+apZlG5r!a|Elv9i%lu9v?>M
zQCoThsrHBvD>v+*UB@re(5QSmr&l1nXxv6lsyr*v?xvNEYl9d)em>p5lT7!EmWfi%
z28-`Ail%DxW0j^F>|w(>3Nh`^Qo>RMRp!7{O{8M~^+`;-B2tud+skTR%ZY78OW3~3
zbL^ZkpOX&BA=-OOOzJ(C$?ta+6>TO%bIUR;k*vkn{8exZ+JGst;dnAA4x@XlM~KNZ
zTy=Nmp41YIot2BjM!`6;Djhu^hC|;@4{H{zK<o2BIA0L(#-=Z<8m7XfZ3z4;%~AF<
z5er87!jku;E4|#XNM$gd_+F-m!9kFobH`f?Lp0qRj^2X|khGrr@i~u)mUly=$pB~`
z;9fW7Xmm_|Mei!U3MOR(;kP0KQx6sK-%}o{I~=G^nh2SUS;u7;yr-iDCX}z(LG4}M
zl5A{&@MYI{GW)4X+N(}jb-B5LPC0xd>DB>cpjje#I!_bY?`;>fwd<*2A!qZyG!dF^
zS~KgCKSEZKz_ja`RpdcO&M_V>)KqP*GTwVc`cl0Y9Vr{bEDltPhMtzB94(a&QOjf*
zW%I>fKKj&ivkdF`qZg|gBV~=V7GlppUsTSC!`@W>Xqt|kLxFJOJ<*P&JTzVBIi`Uk
zJY&<Km28S$eu-$)%!X>Wg-GGK2Nx2<plLf78C+j&+~NU!M;Dy*@Wny101TOyfg{{U
z-WH|8bI}H4&c;ble9%f3oEu}>UkPKUsN(fN?r%9;N7p%9;)v-_sxACO@#Z2~RJ@_S
zYvQRxRK-rYU$nqC7MHJnB8}&YXq{F`5tkas`bs&i?)I2WA_t-}*-_eSs0Od^X(4Q+
z3MT$pPpNUy6zsO$O2Mv@vTo?G)n}iQ_R$b(+%-?|eAHQ}>tw@Z0^`NPY@Xe9Yr56Z
zs&rQTqYrf+BVlI}dNAt`PFBSlw**P&_f>WJnbPVw7j|nrpZnh!AeJ`!GR=p6taXnU
zlRH+yqBV*A;??qbCWqN~p0BI9eLA*AWg>TPDDEsWhU#Vs=B}L0>nL_O;TVFYXV#$R
zO){!o7a}>yA4irRLE&X*IK`!*(A*83{EguLb3C#F=3|WOTpV+7L(<GKusJ*#F2DO@
z<(^lRZ5s=nZDttrmHT+V4u@Mp4EEX#hLZJsnDWfYmKm1ta8!ot%DptjB!krYXV6)b
z5Hj2|5D8ld4l})}&fz1yJnl@hR_H>;I8=Hf&qt^~)D=ag`J}Xxgz+WQskrxQ+GQ|^
zw0*jZ&f4ihMTm+Jt$RSId<`nn+aWfs)T0JdGdlN0pOqF`i;cZCMb825EGDQY)Z)6a
z%hhR2lAU7NJf7znC`}b4Dz~LMaeIaM4^iwyswGo4Zj`EB))gu@HHgkPc39~~I<W9h
zaV)n$g=7xTsxtU2u$ahMSfQ{49ml6(){;Q9b;`qXuPnqXWZ?GVY!nXUbMOl}aH=lC
zg#{7N@byEt#61|#d(m@}vLNs7j^#GNuvn~%?RN&_NWCM9ik#uEq>eWmr(&YRV5Aq@
z^ZtrAq;H4f*)$dIgVM$&rjPldoW=8UA^e_!`^{s~Fm)5TUk2CdN$Jpvx8(gJjSQ!}
zBgvIx^!0rKowrfqS(g2Ax8xoLr%CB;upGvI9)MQK7OU7zGL$iJFuBL-QH?ZK+V?>k
z4X;o{Kjm_vMlPFf+H*$cR|T=O@vJngVZTu7{Z#yQrCQL_GGki9XOhfYW1->r6`|Z}
zxftx<o7KJ#5HeIx2syqgENA0rRyV_5EDv2Ku2`DL8e+c*Ns|{d>#vifx;=-nhe|GF
ze0i>zeKM7W?I{<JduC$g=qya@x(E;L{qg-Y&yRhPiK!K-C@b2){fJXB!Egm?o~GlO
zIzNABnxgY;&gJ2n)3HkY-v0I)?5oJ(el>4s8EGSEj|p1%XX@{^;h5n@Nag+&8;t;z
z?-+*h{<C?eOb}YPx?|pMo&#;y1HJxHK-j!rWacc3sy=rq`k4}5<h`Y6*RgoXpDSc4
z<GR~Oj0_lzqDhSQhH+10O#x~B$Rva5#nhAXsIZ3{26$9cqlyZQCr%)l<9#h^zc~nR
zRGdUtYfDN}`z2bB9S%dTp$Zu3N|sxzY4w7^%swqqOi(&Wbt4Z;b6lp7?KT&(ICNFa
z=Uj$#jYcZ<Oc!btQmx`gIdM&t0l6$&PXXIBnf;1DA;n~~s5QWe=T~}Id7s|KtcP8&
zl(~0Zy3%tS+tHI?wloL56hp9nMh;4hBsf1mnqQ~M=<F?l*8$$ME1r+}HRi}lScstL
zMVNoB0Jk<wN0&#-aH}#8!zNF};T;wjomq&!%CmT0+brbAhM=1h*G`@)gh7EfLceR`
zci9qjKK+%9Z43c>Go0Yuii$DWVBX^~XPO%HmyLwQOm}2;3x&IN1`U(5;^+TfQr)S7
z+_TB_?6U=8vVEu`bt}ERbVf*ev06CYcLp7NTTG2NhM`B$CbEcc6LdB0`TYBVP+s|2
zygK%kU_AP|)He5{c-<_S&Xvhhp|XtVzBz=s@cV?y@w>%wQ-)IeKus*=8o)z7O3eCk
zXRFd-kQxT*k<9wzRs9;wS=O#0Ea+kavqn!^9O}s$cXyH&R(gvK$8U+AuRX<L&L|H*
z<42z@x{^U^1k`G?Fxhr7oNgxr3m4(IlOarMc^y@6BQ#}8k<glnf41}7w*oK3d>Vt6
z{j>4Sb1r_B55-IF5z8B~06%p`qD*x?d;=yz=Oga{tj*&*0CgyT@_?;$IZ7WXBj-Ek
z%XDaC_7i<nDNVtPtphQxsgkOXYC)yzTk2Qng_7|qa5_>*e=cync*akPoPUKzD7m0n
zT?2#8@P12+9%mVN)5X9?q!|*9n!<1@PJT``Rq4{t>t+!8w$PhCV+6JATY~Hvfon_(
zNizPB)uyBCX-w!tF{ilRD&ND1>ebSyr`HC$D$N&~ZYhzqu4rZU(S*HRwN+>|wv+m(
zO=22r<49S35-V$1FVtw<wv?&Lws>B(k7TO4iUwZQf~-b3?HT1qPD56*MTQ0Jevu+G
zKJP6I9>x2mpXZ?8wHTaw76IKLcdQK$#>$2iY*zQf&XQzY51I|K*oY@bGI7ObDN=ur
zM7#s<_m2=Dv&I`kc5xj~vIMu!Y2o!lGgP;3f(ox|Ee_Cu)vHikKR*$(8#2+W$q7G4
zMc_im6Z&2vi|qG(F^XqR%WoQvy_~6ZNFoP~Evs<G%>f%}6xTvI!emb<E`s+`uX3iU
zLOH*W*TxaA?)1HvBlcU}qM@<_==PYNaMu1MG@s~9CLWH|=Tav1x>rX->~{&C3)YI|
zYn}@IaET<GO+6rkXLTP<kQR=x5Vf77tlq@;5ZY@j1dDx}$-2*U!Tx-Tux-X@x^P3C
zrgTc9Y#kRS@t<#bqHd&6C#%dPKKjz~yFY{>SKo@|o#wEzbvfb?$M=HKj%*fR(4Sdf
zP891VOv9kpnaH2B7_mS3n2rjDVl&s^+~zsud~B+@hClq1DU_D^Be0tv)HZn|&4cGj
znug+tUkv*Hia<>sKVO6b81Hn0@1V69@lgg^ykDpMljroba2<cD8!na?A*sL<>DLtS
z`<)Gv=TAUB&y9(kWCFt#I<VO)kBJ{-Fe$<ks<G>-Ri}+wIWP2Y9(fkOzq88nBF`Kj
z)ZBYUK5biw)kM(a>HEm|*JH}ss*jw-p|ooFUP{^CA*4^qr28iyQ|Y?lLbLs2!TOV#
z5H{wCV6mr&%JLV|k;OHFt=cW&&@?xy)9A@RYhF>ljXf>?7){b$il`~uFPdBJqZ*Bw
zR-WZbEdGKlo1&*f8e%9#^j}MAb7d$u){H{*Gg;Q`5tQ9AjkT8C6kX4q5%pA$h`%OV
z;^m<@q-_ty_j8Gm7EeRA5C0r}!?h7D)^P9)MfzxK+^txT&T<A=KFSk;q1jN0a751+
zn>b6y8-|}f(WDcI3R5lIpXz{a$3~)kWE_$+2q(I8?RsD@tXkCtF||tY7(4*xKX~na
zQ4MYB)Cu}?d*bt)7{sSmk_OMR(){_1R)3n0gN-UMl$nW`MaM|@pE8R4G8M94il~0A
z4c#hICb2!1;;YwF?%Quv?IDMNU;H}#>_S%^UJK#eFP{5BhuZse6I%VOg%cY6Maf{z
zDvfxa|E9w0BAjRa#L=Iu(?Y}(M_!BCA8LgY@+P9O@j_CbbkRz3>X9_fD~zhg8M6a+
zXQ<J$H;b>Bz^r4QiCPVMtmf=HY3od$e>Q6gEBE~>lshMg(n|{Lt?3$;|Kg=68yyVa
z@1Zcd7YO+p{#p3b8*eEK`T7QUQe=gwvGXu-T?iiTkHO)PbUfzQY1fWm6zt&hS?(ig
zh#Cxv77*;B12<haeE7ukb&ID!XSprUC$qy7J!|+*n+1#Weh3^i2&=fSWF)`O5?5%!
zp5F)mTq}z{V><B+$j&h5I@CbzKJZVrM#m#R6!rC`jc0h@F0L0gSItB4m*dng)Sn!_
zRgr%5IWkO5pfgu@(Ke${a$XZhg};NS_`DpHdk?1lr)N;4ojQf>>`h9X^<R4;RA>*$
z6mmP1p>=gONz^8h_A3pNR81EPeYXog*kP)<^N%Rw`_#g@=(f<l;FLh}{ppRFHg(9^
zP>sI5sF}4xl#Eoax{x=9DGe_WEYda!rGdLeW5p$+c8;r6(B-Afcf=`SO3N%Lj`2Zt
zl{=ouBq5m3*~4P8xR)ysq+)|XQeICTGaO^Ry>az!AU-YPT3Yu6__R;OKj%4bV%IXL
zzL^SLEfXA99*={qlkhZmB-+M*qAQ(&rXPUTCnebZ?2Q9k?vtErPZW$YfE!bWb=m_u
z{8gR%CAGL0?F*Up7=UwwH4z-YgPOgh)WcvF5;(8UH(wd{8@>rXZWEyNd;krt(#6Ir
zJ>k)JE&pC7dNg7k1%>s**2+ejz3B|e@eQZl=#HjQGW0t?kHTV~iQB4I(u}Ab<ehj(
z@ZB^~s7r2_S}0lym(zNa)|VI>cTbzL4g1r%IbVfFBP((F#dOhS)I^e)RZz|+1s1vD
zsI>ZCC|kX41!*VUvzq$nvzS%Wom$UYk;YF&CbRyn==;xp@p<qn)??U2RCo48t9&ee
zc={k@K>$t{2Ow`t8Wi~4a;M`sJbz=t*`%8g%KQ3hy_ex?L=+yJvPHIv0ha7u%r(I_
z`0+{`Kc;J-KHUc+$Mk_O@ADasaYOt3-_+rykNq!xk;$+T7@OG>burxAm+y?yb9PwP
zqm70v?+V2NJB(OojF_i9Uw6GFTHFV~ZipMI_jQKywNBVpm`7c^mXXYvuYy72T?*PT
z4xd``sQ44#mgyY2m%)!g`ZG!&pF{4uW>H7!8G3Tjm%#m}{?2|BuaQj6QyNJoai^8<
z&Y5E5v5SI+uN@s~7U*hdrI0(-jA~0iiZ$oss3BA$RLt}w-;)vIOT9LsbuOouY7Y`u
zKi@9u?PwP@mnHDbA2psoJ(_Wjs5E%%e9>3FQrz!j#F|nRS-Kd9ag~I$nG3KiIfL_B
zbW!EE4ihGXfUyY9Yw?1?8Ls^r7zcJtAMc{&(Hd)y&h_Krv@acx#yjIo6t4-YDsug@
z6Vfi~A-@x6@y^tN-nl^B`W%9e+d6per;Aq`I9qWi_jX0d!}8Gyl5hr-**jTOmbTKs
zcS<n+rVOXnFZ4Qx^P>;)I<jUxwV1u5sX0G|?{hM#&hZT0Q>dmVa+>sR!bMuTUZB>C
zjY9U_uJpuKo%6jp>+{Px^3^q>Y9|NMYR#n2ZOua2r%^l$eTeAd)*1B%nS$X$8(Mni
zu3(tc8E<loN&fL7p<&4}(et#CnBP;MB=JM7oF{3r=-+ENM|GF@(q%D8COoc^;Vch}
z<~mVETZJjRC^F|c%G`IlmI8RDi1OqNG5T>=CV6~E)a_%5BCjx<<7|wum;64-eiMuz
zTcY?$Ac9(F;z?@+jx|}LVDn(SOfg5q!euyhjr$L;j^?$}anLjwf)e?qyfz(!-@T?{
z)@xv9ye;zV5|Cij4J=?Nvd{OzqF5`8<Kxla;w@z_2;?4cO(=Gl;*<Skv<!Vp+qwTC
z(EB2_MRAs$wmhCb*+{wCz9?g_=yjP1&Kx(S;VCz`PcMb47oH}k?ut+|SAfq36S|&i
zOS{q^k*e<=+N5)y3WGY~<(iX%lA9v+yIf8Z`A|{IBbtu++@MC?nO3zsrwTIWt(J{<
z28uS_OsP0ZiE7{X7Od3=h}KUc1h0!XNk+G?@Zs@93Xhz?8b^K-mQHgJ*4P~+-;H(B
zVT*V+Wn=fM>l#{AJYu^T)=idsq<>j<7;R(^Ukzm9gvG#(5EOJa#pO{m@${esQH~Qa
zHzNVtxW`YU$dWTiC&A{Q0btt_u`MVH8^^k0^J5qM=*4>&^8UQWstFBGU98|*ifg>C
zY-`^K8T0z$<LnU#&{4)2W`SS-DC5$@Skz^yAm!M7@>bBoEZ5PP!n05p-|CLWYw|GK
z^PUPjIiXN_C*2;!8C9x1v2l+c0-BGKOeag2nKseeg<Q|GzLnmnETZ8JwbXV&4c;-n
z^rMdlwXV4*I9JvR&+D&~Q)@Q_{!`3*dn?2}J&LJ*@jk)gkG{})<cFYoK3D8jwU$zT
zW{4hj+o*A%HkI_zrq)g}qW7m)!hU^AvS=%&UqRail`mE_;(8vrb~!0Zw#!?V9lj@?
z3+heA`)kBE9--oLi5?p!_Mpgp1DI2qDl^HlWd8G=kPu`By$`&{Up@(%<FXJp)E5sX
zC}8s*Cu~dM+Js-O82C_vlr<hG?&6MUJw=pg&c`^ecet_H4n-{{xIq25r>8GkIZtk8
z$ZV)Knc`);F~Zi#;Okm_$g9Rdc{*oF-0q80m2N2A<&G)+Eb--j3)vh~#!k0Kbaeex
z91;`}IoTE(k^3mzK9Y_;zeMiS2J(JE3iXa&OYa&7qLgc=MowN!H}ucZkK%FE&NGi+
zG&a!(e-|p-FqC{3i$eQ`rF1iK6^-S(V97+;s^_CBD7VjQ(zLormA2ZXTJ%C(9bYXd
z4-|yid-X`R*F)jmo^4cmjiiO6dXeO+Syjq_Bf@%TEs{LUv6OSuBh~!7qH;y4Xl>Ak
zEl>U_x>b~uifb;7xp!Mg>lV!RydTRFyK#@TV>fyIxt9BK&UTXd`*3#bA}b>!FE8^y
z7IhhYnV4w5@I?_}ivy$QM#RQM#>N=&r;8RugxlE+m6`qjvzRLU-QvxX`EH&07XCi|
z-&@p^(fb!$Gz{>I@iST!7`?zRY(czV4Bz_P1p#(8L;k;Zz<8{=nZ;NOi?QaW7Go_f
z%}mTKOpN(5+QQh(!ra8tbnIwLGt;qS|L*1C5pP%N@jd+eaCYqUcW3`&WJY`Z^P_^l
z{`=7Q`yc;h`JdzTcc%>_L&N_2dQ$O1U5@|k-^YOevW{Q;)&E?Ne_TvpM07x4^nYK`
ze?Ew6|7At)o&ME|{`V*v|M#{0=L4huU)JJ#zSF<hoqs?4|9v(8dF1|%-v3>0{ma$-
z>tpKdC?{v|??0;5Wz>JHr0c&w`#=B2@L%xnvh>2gJ>fss?eFX6i~sie@88ajUH=ZT
OOjrKr@B9Dlz5few;fy~3

diff --git a/test/models/test_transolver.py b/test/models/test_transolver.py
index 8b29e78e82..b6775f7f61 100644
--- a/test/models/test_transolver.py
+++ b/test/models/test_transolver.py
@@ -18,19 +18,29 @@
 
 import pytest
 import torch
+from common import (
+    check_ort_version,
+    validate_amp,
+    validate_checkpoint,
+    validate_combo_optims,
+    validate_cuda_graphs,
+    validate_forward_accuracy,
+    validate_jit,
+    validate_onnx_export,
+    validate_onnx_runtime,
+)
+from pytest_utils import import_or_fail
 
 from physicsnemo.models.transolver import Transolver
 
-from . import common
-
 
 @pytest.mark.parametrize("device", ["cuda:0", "cpu"])
-def test_transolver_forward(device):
-    """Test FNO forward pass"""
+def test_transolver2d_forward(device):
+    """Test Transolver2D forward pass"""
     torch.manual_seed(0)
-    # Construct FNO model
+    # Construct Transolver model
     model = Transolver(
-        space_dim=2,
+        structured_shape=(85, 85),
         n_layers=8,
         n_hidden=64,
         dropout=0,
@@ -38,36 +48,42 @@ def test_transolver_forward(device):
         Time_Input=False,
         act="gelu",
         mlp_ratio=1,
-        fun_dim=1,
+        functional_dim=1,
         out_dim=1,
         slice_num=32,
-        ref=8,
-        unified_pos=1,
-        H=85,
-        W=85,
+        ref=1,
+        unified_pos=True,
+        use_te=False,
     ).to(device)
 
     bsize = 4
-    pos = torch.randn(bsize, 85, 85).to(device)
-    invar = torch.randn(bsize, 85 * 85, 1).to(device)
 
-    assert common.validate_forward_accuracy(
+    fx = torch.randn(bsize, 85 * 85, 1).to(device)
+    embedding = torch.randn(bsize, 85, 85).to(device)
+
+    print(f"fx: {fx.shape}")
+    print(f"embedding: {embedding.shape}")
+
+    print(f"output shape: {model(fx, embedding).shape}")
+
+    assert validate_forward_accuracy(
         model,
         (
-            pos,
-            invar,
+            fx,
+            embedding,
         ),
-        file_name="transolver_output.pth",
+        file_name="transolver2d_output.pth",
         atol=1e-3,
     )
 
 
 @pytest.mark.parametrize("device", ["cuda:0", "cpu"])
-def test_transolver_constructor(device):
-    """Test transolver constructor options"""
-    # Define dictionary of constructor args
+def test_transolver_irregular_forward(device):
+    """Test Transolver Irregular forward pass"""
+    torch.manual_seed(0)
+    # Construct Transolver model
     model = Transolver(
-        space_dim=2,
+        structured_shape=None,
         n_layers=8,
         n_hidden=64,
         dropout=0,
@@ -75,31 +91,40 @@ def test_transolver_constructor(device):
         Time_Input=False,
         act="gelu",
         mlp_ratio=1,
-        fun_dim=1,
+        functional_dim=2,
+        embedding_dim=3,
         out_dim=1,
         slice_num=32,
-        ref=8,
-        unified_pos=1,
-        H=85,
-        W=85,
+        ref=1,
+        unified_pos=False,
+        use_te=False,
     ).to(device)
 
-    bsize = random.randint(1, 4)
-    pos = torch.randn(bsize, 85, 85).to(device)
-    invar = torch.randn(bsize, 85 * 85, 1).to(device)
+    bsize = 4
+
+    embedding = torch.randn(bsize, 12345, 3).to(device)
+    functional_input = torch.randn(bsize, 12345, 2).to(device)
 
-    outvar = model(pos, invar)
-    assert outvar.shape == (bsize, 85, 85, 1)
+    assert validate_forward_accuracy(
+        model,
+        (
+            embedding,
+            functional_input,
+        ),
+        file_name="transolver_irregular_output.pth",
+        atol=1e-3,
+    )
 
 
-@pytest.mark.parametrize("device", ["cuda:0", "cpu"])
+@pytest.mark.parametrize("device", ["cuda:0"])
 def test_transolver_optims(device):
     """Test transolver optimizations"""
 
     def setup_model():
         """Setups up fresh transolver model and inputs for each optim test"""
+
         model = Transolver(
-            space_dim=2,
+            structured_shape=None,
             n_layers=8,
             n_hidden=64,
             dropout=0,
@@ -107,24 +132,25 @@ def setup_model():
             Time_Input=False,
             act="gelu",
             mlp_ratio=1,
-            fun_dim=1,
+            functional_dim=2,
+            embedding_dim=3,
             out_dim=1,
             slice_num=32,
-            ref=8,
-            unified_pos=1,
-            H=85,
-            W=85,
+            ref=1,
+            unified_pos=False,
+            use_te=False,
         ).to(device)
 
-        bsize = random.randint(1, 2)
-        pos = torch.randn(bsize, 85, 85).to(device)
-        invar = torch.randn(bsize, 85 * 85, 1).to(device)
+        bsize = 4
+
+        embedding = torch.randn(bsize, 12345, 3).to(device)
+        functional_input = torch.randn(bsize, 12345, 2).to(device)
 
-        return model, pos, invar
+        return model, embedding, functional_input
 
     # Ideally always check graphs first
     model, pos, invar = setup_model()
-    assert common.validate_cuda_graphs(
+    assert validate_cuda_graphs(
         model,
         (
             pos,
@@ -134,7 +160,7 @@ def setup_model():
 
     # Check JIT
     model, pos, invar = setup_model()
-    assert common.validate_jit(
+    assert validate_jit(
         model,
         (
             pos,
@@ -143,7 +169,7 @@ def setup_model():
     )
     # Check AMP
     model, pos, invar = setup_model()
-    assert common.validate_amp(
+    assert validate_amp(
         model,
         (
             pos,
@@ -152,7 +178,7 @@ def setup_model():
     )
     # Check Combo
     model, pos, invar = setup_model()
-    assert common.validate_combo_optims(
+    assert validate_combo_optims(
         model,
         (
             pos,
@@ -161,12 +187,53 @@ def setup_model():
     )
 
 
+@import_or_fail("transformer_engine")
+def test_transolver_te(pytestconfig):
+    if not torch.cuda.is_available():
+        pytest.skip("CUDA is not available")
+
+    torch.manual_seed(0)
+
+    model = Transolver(
+        structured_shape=None,
+        n_layers=8,
+        n_hidden=64,
+        dropout=0,
+        n_head=4,
+        Time_Input=False,
+        act="gelu",
+        mlp_ratio=1,
+        functional_dim=2,
+        embedding_dim=3,
+        out_dim=1,
+        slice_num=32,
+        ref=1,
+        unified_pos=False,
+        use_te=True,
+    ).to("cuda")
+
+    bsize = 4
+
+    embedding = torch.randn(bsize, 12345, 3).to("cuda")
+    functional_input = torch.randn(bsize, 12345, 2).to("cuda")
+
+    assert validate_forward_accuracy(
+        model,
+        (
+            embedding,
+            functional_input,
+        ),
+        file_name="transolver_irregular_te_output.pth",
+        atol=1e-3,
+    )
+
+
 @pytest.mark.parametrize("device", ["cuda:0", "cpu"])
 def test_transolver_checkpoint(device):
     """Test transolver checkpoint save/load"""
     # Construct transolver models
     model_1 = Transolver(
-        space_dim=2,
+        structured_shape=None,
         n_layers=8,
         n_hidden=64,
         dropout=0,
@@ -174,17 +241,17 @@ def test_transolver_checkpoint(device):
         Time_Input=False,
         act="gelu",
         mlp_ratio=1,
-        fun_dim=1,
+        functional_dim=2,
+        embedding_dim=3,
         out_dim=1,
         slice_num=32,
-        ref=8,
-        unified_pos=1,
-        H=85,
-        W=85,
+        ref=1,
+        unified_pos=False,
+        use_te=False,
     ).to(device)
 
     model_2 = Transolver(
-        space_dim=2,
+        structured_shape=None,
         n_layers=8,
         n_hidden=64,
         dropout=0,
@@ -192,36 +259,37 @@ def test_transolver_checkpoint(device):
         Time_Input=False,
         act="gelu",
         mlp_ratio=1,
-        fun_dim=1,
+        functional_dim=2,
+        embedding_dim=3,
         out_dim=1,
         slice_num=32,
-        ref=8,
-        unified_pos=1,
-        H=85,
-        W=85,
+        ref=1,
+        unified_pos=False,
+        use_te=False,
     ).to(device)
 
     bsize = random.randint(1, 2)
-    pos = torch.randn(bsize, 85, 85).to(device)
-    invar = torch.randn(bsize, 85 * 85, 1).to(device)
 
-    assert common.validate_checkpoint(
+    embedding = torch.randn(bsize, 12345, 3).to(device)
+    functional_input = torch.randn(bsize, 12345, 2).to(device)
+
+    assert validate_checkpoint(
         model_1,
         model_2,
         (
-            pos,
-            invar,
+            functional_input,
+            embedding,
         ),
     )
 
 
-@common.check_ort_version()
+@check_ort_version()
 @pytest.mark.parametrize("device", ["cuda:0", "cpu"])
-def test_transolverdeploy(device):
+def test_transolver_deploy(device):
     """Test transolver deployment support"""
     # Construct transolver model
     model = Transolver(
-        space_dim=2,
+        structured_shape=(85, 85),
         n_layers=8,
         n_hidden=64,
         dropout=0,
@@ -229,27 +297,27 @@ def test_transolverdeploy(device):
         Time_Input=False,
         act="gelu",
         mlp_ratio=1,
-        fun_dim=1,
+        functional_dim=1,
         out_dim=1,
         slice_num=32,
-        ref=8,
-        unified_pos=1,
-        H=85,
-        W=85,
+        ref=1,
+        unified_pos=True,
+        use_te=False,
     ).to(device)
 
-    bsize = random.randint(1, 2)
-    pos = torch.randn(bsize, 85, 85).to(device)
-    invar = torch.randn(bsize, 85 * 85, 1).to(device)
+    bsize = 4
+
+    pos = torch.randn(bsize, 85 * 85, 1).to(device)
+    invar = torch.randn(bsize, 85, 85).to(device)
 
-    assert common.validate_onnx_export(
+    assert validate_onnx_export(
         model,
         (
             pos,
             invar,
         ),
     )
-    assert common.validate_onnx_runtime(
+    assert validate_onnx_runtime(
         model,
         (
             invar,