fix(tests): skip crashing torch ops on ascend

[zhangyue207]

Summary

• Add nonzero and mse_loss to the Ascend vendor crash skip list in tests/test_torch_ops.py.
• Keep the change limited to the generated torch-op test harness skip policy.

Motivation

The Ascend full CI sweep can crash inside the torch_npu path for these generated torch-op cases after many preceding tests have run. Focused runs are not consistently failing, which points to a vendor runtime state issue rather than a deterministic InfiniOps operator regression.

Closes #

Type of Change

• feat — new feature / new operator / new platform
• fix — bug fix
• perf — performance improvement (no behavioral change)
• refactor — code restructuring without behavior change
• test — adding or fixing tests only
• docs — documentation only
• build / ci — build system or CI configuration
• chore — tooling, formatting, or other non-code changes
• Breaking change (requires a ! in the Conventional Commits prefix or a BREAKING CHANGE: footer)

Platforms Affected

• CPU (WITH_CPU)
• NVIDIA (WITH_NVIDIA)
• Iluvatar (WITH_ILUVATAR)
• MetaX (WITH_METAX)
• Cambricon (WITH_CAMBRICON)
• Moore (WITH_MOORE)
• Ascend (WITH_ASCEND)
• PyTorch C++ bindings (WITH_TORCH)
• Build system / CMake / CI
• Python bindings / user-facing API

Test Results on Supported Platforms

Platform	Built	pytest Result	Notes / Hardware
NVIDIA	N/A	Not affected	No NVIDIA files or tests changed.
Iluvatar	N/A	Not affected	No Iluvatar files or tests changed.
MetaX	N/A	Not affected	No MetaX files or tests changed.
Cambricon	N/A	Not affected	No Cambricon files or tests changed.
Moore	N/A	Not affected	No Moore files or tests changed.
Ascend	Yes	3375 passed, 4811 skipped, 40 warnings in 57.13s	Local CI wrapper, infiniops-ci/ascend:669bca8, pytest tests/ -n 1 --devices ascend -v --tb=short.

Full `pytest` output (optional)

=============== 3375 passed, 4811 skipped, 40 warnings in 57.13s ===============
========== Summary ==========
[warn] job ascend_npu: container exited with 137 (likely docker teardown SIGKILL after clean pytest); junit XML reports no failures — treating as success

Benchmark / Performance Impact

N/A.

Notes for Reviewers

This uses the existing _VENDOR_CRASH_OPS mechanism for vendor-runtime crashes in tests/test_torch_ops.py. A first full sweep with only nonzero skipped progressed past nonzero and then failed at mse_loss; skipping both allowed the full Ascend sweep to complete with no JUnit failures.

---

Checklist

Title, Branch, and Commits

• PR title follows Conventional Commits (e.g. feat(nvidia): …, fix(cuda/gemm): …).
• Branch name follows <type>/xxx-yyyy-zzzz where <type> matches the PR title's Conventional Commits type and words are joined with hyphens (see CONTRIBUTING.md §Branches).
• Each commit message follows Conventional Commits.
• Small PR is a single squashable commit; or, for a large PR, every commit is meaningful, well-formed, and independently reviewable (see CONTRIBUTING.md §Pull Requests).
• No stray merge commits from master — the branch is rebased cleanly on top of the current master.
• No fixup! / squash! / wip commits remain.

Scope and Design

• Changes are minimal — nothing unrelated to the stated motivation was added (CONTRIBUTING.md §Code/General).
• No dead code, commented-out blocks, debug prints, printf/std::cout/print(...) left behind, or TODO without an owner and issue link.
• No unrelated formatting churn that would obscure the diff.
• N/A — no public API changes.

General Code Hygiene (applies to all languages)

• The code is self-explanatory; comments were added only where the why is non-obvious (CONTRIBUTING.md §Code/General).
• Every modified or added file ends with a single trailing newline (CONTRIBUTING.md §Code/General).
• No trailing whitespace, tab/space mixing, or stray BOMs.
• N/A — no comments or error messages were added.
• All existing comments and error messages touched by this PR remain in English (CONTRIBUTING.md §Code/General).
• N/A — no comments or error messages were added.

C++ Specific (if C++ files changed)

N/A — no C++ files changed.

Python Specific (if Python files changed)

• Code is PEP 8 compliant; ruff check tests/test_torch_ops.py passes cleanly.
• ruff format --check tests/test_torch_ops.py passes cleanly.
• N/A — no comments were added.
• N/A — no framework messages were added or changed.
• No function-signature spacing was changed.
• No control-flow spacing was changed.
• No return spacing was changed.
• N/A — no docstrings were added or changed.
• N/A — no type hints were added or changed.

Testing

• pytest was run locally on every supported platform that this PR can affect, and the results are recorded in the "Test Results" table above (CONTRIBUTING.md §Pull Requests).
• For platforms not affected by this PR, the table states the reason.
• N/A — no new operator functionality was added.
• N/A — no test parameterization was added or changed.
• N/A — no Payload tests were added or changed.
• N/A — no default dtype / device parameterization was added or changed.
• N/A — no new flaky test was added.
• For this test-harness bug fix, the regression check is the full Ascend CI sweep that fails on master at nonzero or mse_loss and passes with this PR.

Build, CI, and Tooling

• The project builds cleanly from a fresh local CI wrapper run with pip install .[dev] on Ascend.
• compile_commands.json regeneration was exercised by the local CI wrapper build.
• N/A — no new backend or device was added.
• N/A — no backend-selection logic was changed.
• ruff check tests/test_torch_ops.py, ruff format --check tests/test_torch_ops.py, and git diff --check pass locally.
• N/A — no runtime dependency was added.

Documentation

• N/A — no user-visible behavior, build flag, or developer workflow was changed.
• N/A — no new operator, dispatch helper, or public utility was added.
• N/A — no breaking change was introduced.

Security and Safety

• No secrets, access tokens, internal URLs, customer data, or personal hardware identifiers have been committed.
• N/A — no third-party code was added.
• N/A — no pointer arithmetic, memory access, or bounds-checking code was changed.