[Flex Attention] Remove forked pytorch usage from benchmarks workflow

[vlad-penkin]

intel-xpu-backend-for-triton/.github/workflows/triton-benchmarks.yml

Line 287 in d9e0891

# Install Pytorch with FlexAttention XPU support enabled

Our guiding principle is to use upstream PyTorch repo only if some PR's are not merged add those here:

• https://github.com/intel/intel-xpu-backend-for-triton/blob/main/scripts/patch-pytorch.sh