Test changes to linux_job_v2.yml via regression_test_rocm.yml

Test changes to linux_job_v2.yml via regression_test_rocm.yml

Support _weight_int8pack_mm path for all backends and INT8 quantizati…

…on on Intel XPU

updates

Update workflow to use amdfaa/test-infra

use python version agnostic binding for mxfp8 cuda kernels (pytorch#3471

)

* use py agnostic c++ extension for mxfp8_cuda

* refactor mxfp8 cuda from pybind to torch_library api

* put schema def inside guard

use python version agnostic binding for mxfp8 cuda kernels (pytorch#3471

)

* use py agnostic c++ extension for mxfp8_cuda

* refactor mxfp8 cuda from pybind to torch_library api

* put schema def inside guard

update version compatibility table (pytorch#3455)

Update quantization config for small_bf16_linear model

Support mx_tensor and enable it's test on Intel GPU