[E2E] Fix HUGGING_FACE_HUB_TOKEN usage for e2e perf (#5461)

Continuation of #5399.
Enabling the same for the e2e performance workflows.
Passing CI:
https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/19277885211

Author: dev-tomek

.github/workflows/e2e-performance.yml

@@ -128,6 +128,8 @@ jobs:
         dtype: ${{ fromJson(needs.setup.outputs.dtype) }}
       fail-fast: false
     uses: ./.github/workflows/e2e-reusable.yml
+    secrets:
+      HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
     with:
       pytorch_ref: ${{ inputs.pytorch_ref }}
       suite: ${{ matrix.suite }}