perf(l1): improve rlp encoding (#5350)

**Motivation**

- In the current impl_encode there are 2 loops to get the position,
fixed to a single pass.
- Added specialized branchless length methods to integers,
- Added specialized length to Vec<T>
- Added specialized length to tuples
- Added specialized length to most types in ethrex-common, only ones
missing are in networking/p2p
- Removed unneeded arrayvec usage
- Many instances that deref to &[u8] didn't use its length specialized
method, fixed to use it

Each missing length impl in the trait impl meant an extra allocation,
this should remove them entirely when encoding

Helps with #4956

---------

Co-authored-by: Tomás Grüner <47506558+MegaRedHand@users.noreply.github.com>

Author: edg-l

CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## Perf
 
+### 2025-11-20
+
+- Improve rlp encoding by avoiding extra loops and remove unneeded array vec, also adding a alloc-less length method the default trait impl [#5350](https://github.com/lambdaclass/ethrex/pull/5350)
+
 ### 2025-11-19
 
 - Parallelize merkleization [#5377](https://github.com/lambdaclass/ethrex/pull/5377)