Make parquet memory tests opt-in

[alexbarghi-nv]

Summary

• backport Make parquet memory tests opt-in #535 to release/26.10
• skip Parquet peak host-memory scaling tests by default because RSS varies nondeterministically across CI environments
• keep both tests available for local investigation with PYLIBWHOLEGRAPH_RUN_PARQUET_MEMORY_TESTS=1
• leave functional Parquet I/O coverage unchanged

Testing

• git diff --check
• PYTHONPYCACHEPREFIX=/tmp/cugraph-gnn-pr535-release-pycache python3 -m py_compile python/pylibwholegraph/pylibwholegraph/tests/pylibwholegraph/test_tensor_file_formats_io.py

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[greptile-apps]

Greptile code reviews should be reviewed for accuracy and completeness. Please remember to give Greptile feedback (👍/👎) and consider leaving comments explaining why Greptile's feedback was helpful or not. This will help improve future code reviews.

Confidence Score: 5/5

[Low risk] Makes parquet memory tests skip by default.

The PR appears safe to merge.

Summary

This backport makes both Parquet peak-memory tests opt-in while leaving functional Parquet I/O tests enabled. The change since the previous review is formatting only.

_{Reviews (2) · Last reviewed commit: "Apply ruff formatting"}

[alexbarghi-nv]

/ok to test 011a797

[alexbarghi-nv]

/ok to test 0e3402c