[CI/Build] Fix disabled v1 attention backend selection test

[Isotr0py]

Purpose

• Related to [CI]: Tests skipped for V0 deprecation #25364
• Fix disabled test_env test in tests/kernels/attention/test_attention_selector.py

Test Plan

pytest -s -v tests/kernels/attention/test_attention_selector.py -k test_env

Test Result

All tests should pass now:

tests/kernels/attention/test_attention_selector.py::test_env[cuda_TRITON_MLA_mla_T_blks16] INFO 09-23 19:32:46 [cuda.py:278] Using Triton MLA backend on V1 engine.
W0923 19:32:46.830000 825531 .venv/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2425] TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
W0923 19:32:46.830000 825531 .venv/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2425] If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'] to specific architectures.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_TRITON_MLA_mla_T_blks64] PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASHMLA_mla_T_blks16] SKIPPED (FlashMLA only supports block_size 64)
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASHMLA_mla_T_blks64] SKIPPED (FlashMLA not supported on this platform)
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASHINFER_MLA_mla_T_blks16] SKIPPED (FlashInfer MLA only supports block_size 32 or 64)
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASHINFER_MLA_mla_T_blks64] INFO 09-23 19:32:48 [cuda.py:259] Using FlashInfer MLA backend on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASH_ATTN_MLA_mla_T_blks16] INFO 09-23 19:32:48 [cuda.py:273] Using FlashAttention MLA backend on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASH_ATTN_MLA_mla_T_blks64] PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_CUTLASS_MLA_mla_T_blks16] SKIPPED (CUTLASS_MLA only supports block_size 128)
tests/kernels/attention/test_attention_selector.py::test_env[cuda_CUTLASS_MLA_mla_T_blks64] SKIPPED (CUTLASS_MLA only supports block_size 128)
tests/kernels/attention/test_attention_selector.py::test_env[hip_TRITON_MLA_mla_T_blks16] INFO 09-23 19:32:49 [rocm.py:209] Using Triton MLA backend on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[hip_TRITON_MLA_mla_T_blks1] PASSED
tests/kernels/attention/test_attention_selector.py::test_env[hip_ROCM_AITER_MLA_mla_T_blks16] PASSED
tests/kernels/attention/test_attention_selector.py::test_env[hip_ROCM_AITER_MLA_mla_T_blks1] INFO 09-23 19:32:50 [rocm.py:218] Using AITER MLA backend on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_XFORMERS_mla_F_blks16] INFO 09-23 19:32:50 [cuda.py:347] Using Flash Attention backend on V1 engine.
INFO 09-23 19:32:50 [cuda.py:364] Using FlexAttention backend for head_size=16 on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cuda_FLASHINFER_mla_F_blks16] INFO 09-23 19:32:51 [cuda.py:293] Using FlashInfer backend on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[hip_ROCM_FLASH_mla_F_blks16] INFO 09-23 19:32:51 [rocm.py:245] Using Triton Attention backend on V1 engine.
PASSED
tests/kernels/attention/test_attention_selector.py::test_env[cpu_TORCH_SDPA_mla_F_blks16] INFO 09-23 19:32:52 [cpu.py:101] Using Torch SDPA backend.
PASSED

================================================================================ warnings summary =================================================================================
.venv/lib/python3.12/site-packages/torch/cuda/__init__.py:63
  /home/mozf/develop-projects/vllm/.venv/lib/python3.12/site-packages/torch/cuda/__init__.py:63: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
    import pynvml  # type: ignore[import]

.venv/lib/python3.12/site-packages/schemathesis/generation/coverage.py:305
  /home/mozf/develop-projects/vllm/.venv/lib/python3.12/site-packages/schemathesis/generation/coverage.py:305: DeprecationWarning: jsonschema.exceptions.RefResolutionError is deprecated as of version 4.18.0. If you wish to catch potential reference resolution errors, directly catch referencing.exceptions.Unresolvable.
    ref_error: type[Exception] = jsonschema.RefResolutionError,

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
============================================================= 13 passed, 5 skipped, 4 deselected, 2 warnings in 6.33s =============================================================

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.

[gemini-code-assist]

Code Review

This pull request re-enables a disabled test for the v1 attention backend selection and corrects an import path within it. The changes are straightforward and correct, successfully fixing the test as described. I find no issues with this update.