Record accuracy sweep results for bilateral

[brendancol]

Summary

Records accuracy sweep results for the bilateral module in .claude/sweep-accuracy-state.csv. No issues found.

Audit covered all five accuracy categories across numpy, cupy, dask+numpy, and dask+cupy backends:

• Floating point: float64 used throughout. Sigma underflow already validated via sqrt(np.finfo(float64).tiny) lower bound; oversize sigma clamped to max(rows, cols).
• NaN handling: NaN center returns NaN; NaN neighbors are skipped without incrementing the denominator. The w_sum > 0.0 guard prevents division by zero.
• Off-by-one: map_overlap depth equals kernel radius. CUDA bounds use strict <. Loop bounds are symmetric and inclusive of y+radius.
• Earth curvature: not applicable (pixel space).
• Backend consistency: numpy and cupy paths use identical algorithms with float64 dtype. Dask paths route through the same chunk function.

Test plan

• Read xrspatial/bilateral.py, xrspatial/utils.py, xrspatial/tests/general_checks.py, xrspatial/tests/test_bilateral.py
• CSV update visible in diff

[brendancol]

@copilot resolve the merge conflicts in this pull request

[Copilot]

@copilot resolve the merge conflicts in this pull request

Merged origin/main into this branch in a126bbe. The merge resolved automatically with no conflicts — the CSV update on this branch was cleanly integrated with the upstream changes to glcm.py, sky_view_factor.py, and their tests.