[dannyward630]

Fixes #3752

Summary

• Treat the documented non-affine GCP NotImplementedError as an expected refusal in single-byte mutation fuzzing.
• Keep the valid-by-construction IFD layout test limited to ValueError and TypeError.
• Add a deterministic regression for the reported le_geo_f32 mutation at byte 146.

The reader's NotImplementedError is intentional for non-affine GCP tiepoints. This change keeps reader behavior intact and narrows the fuzz allowance to that specific GCP-warp error.

Validation

• python -m pytest xrspatial/geotiff/tests/test_fuzz_hypothesis.py -q — 16 passed.
• Focused mutation regression and property cases — 5 passed.
• flake8 and isort --check-only on the changed file — passed.

AI-assisted review notes

The xarray-spatial review checklist was applied to correctness, error handling, scope, and test coverage.

Issues found:

• A global exception allow-list would weaken the valid-by-construction parser test, so NotImplementedError is limited to mutation fuzzing and matched to the documented GCP-warp message.
• The exact reported byte mutation is now exercised deterministically.

Open concerns:

• None identified in this test-only change.

[brendancol]

@dannyward630 hey hey...beep beep boop boop...I loved your gh profile description!

I'm review ASAP and will get merged and released. Thanks again!

[brendancol]

@dannyward630 hey would you mind filling out a new contributor issue?

Reviewed and going to reject for not following AI guidelines. Happy to re-review once you have filled out a new contributor issue.

Overall, not sure how much contribution across github is too much, but 300+ PRs across 200+ repos in a short period of time is too many to not follow xarray-spatial guidelines.