[tests] enable awq related tests on XPU

[faaany]

What does this PR do?

As the title suggests, this PR makes awq tests that only run CUDA to run on XPU. Below is an evidence:

============================================================================= short test summary info =============================================================================
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[128-128]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[128-256]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[128-512]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[128-1024]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[256-128]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[256-256]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[256-512]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[256-1024]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[512-128]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[512-256]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[512-512]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[512-1024]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[1024-128]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[1024-256]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[1024-512]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_from_qbits_tensor[1024-1024]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_move[cpu]
PASSED test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_move[xpu]
SKIPPED [400] test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py:115: CUDA >= sm80 not available
=================================================================== 18 passed, 400 skipped, 1 warning in 1.33s ====================================================================

[dacorvo]

Thanks for this pull-request ! A few chnges are required to align with the existing test pattern.

[SunMarc]

Looks much better, thanks !

[dacorvo]

Almost there. You also need to use meaningful commits (at least a small sentence) otherwise they will be rejected by the CI. Your commit that contain the single word "update" is currently rejected. You can amend it and force-push (push -f) your branch.

[faaany]

done, thx!

[faaany]

I fixed the failed tests, but this test "test/tensor/weights/optimized/test_awq_weight_qbits_tensor.py::test_awq_weight_qbits_tensor_move[cuda]" still fails, also in main branch.