test: add maximum operator coverage

[voltjia]

Summary

• Adds handwritten maximum operator coverage in tests/test_maximum.py.
• Covers contiguous and strided inputs/outputs across float32, float16, and bfloat16.

Motivation

Generated smoke coverage confirms the PyTorch backend path is wired, but it does not provide focused per-operator correctness coverage. This PR adds a dedicated maximum test following the existing operator-test style.

Closes N/A

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

Smoke Test Result

ruff format --check tests/test_maximum.py tests/test_minimum.py tests/test_addcmul.py tests/test_addcdiv.py
4 files already formatted

ruff check tests/test_maximum.py tests/test_minimum.py tests/test_addcmul.py tests/test_addcdiv.py
All checks passed!

Cross-platform cropped smoke validation:
python3 -m pytest tests/test_maximum.py tests/test_minimum.py tests/test_addcmul.py tests/test_addcdiv.py -q -v --devices <cpu + platform>
138 passed, 6 skipped

Test Results on Supported Platforms

Platform	Affected	Build / Smoke Result	Full Result / Notes
NVIDIA	Yes	passed; build 158s, test 7s, total 165s	138 passed, 6 skipped
Iluvatar	Yes	passed; build 149s, test 4s, total 153s	138 passed, 6 skipped
MetaX	Yes	passed; build 242s, test 15s, total 258s	138 passed, 6 skipped
Cambricon	Yes	passed; build 338s, test 10s, total 348s	138 passed, 6 skipped
Moore	Yes	passed; build 336s, test 21s, total 357s	138 passed, 6 skipped
Ascend	Yes	pytest passed; build 195s, test 14s, total 209s	138 passed, 6 skipped; container exit code was 137 after pytest summary, consistent with prior Ascend runs

Full `pytest` output (optional)

NVIDIA:    tests/test_maximum.py + tests/test_minimum.py + tests/test_addcmul.py + tests/test_addcdiv.py: 138 passed, 6 skipped in 2.58s
Iluvatar:  tests/test_maximum.py + tests/test_minimum.py + tests/test_addcmul.py + tests/test_addcdiv.py: 138 passed, 6 skipped in 0.80s
MetaX:     tests/test_maximum.py + tests/test_minimum.py + tests/test_addcmul.py + tests/test_addcdiv.py: 138 passed, 6 skipped in 0.71s
Cambricon: tests/test_maximum.py + tests/test_minimum.py + tests/test_addcmul.py + tests/test_addcdiv.py: 138 passed, 6 skipped in 3.93s
Moore:     tests/test_maximum.py + tests/test_minimum.py + tests/test_addcmul.py + tests/test_addcdiv.py: 138 passed, 6 skipped in 14.23s
Ascend:    tests/test_maximum.py + tests/test_minimum.py + tests/test_addcmul.py + tests/test_addcdiv.py: 138 passed, 6 skipped in 2.27s

Benchmark / Performance Impact

N/A - test-only change.

Notes for Reviewers

The six skipped cases are from the batch companion addcdiv test: PyTorch addcdiv.out does not support float16 on CPU, so those CPU/PyTorch-backend cases are skipped explicitly.