polygonize: close test-coverage gaps for backends, edge cases, params

[brendancol]

Summary

Deep-sweep test-coverage pass against xrspatial/polygonize.py on 2026-05-19. Adds xrspatial/tests/test_polygonize_coverage_2026_05_19.py with 58 tests, all passing on a CUDA host.

Gaps closed

Cat 1 (backend coverage)

• simplify_tolerance= and mask= parity on the dask+cupy backend. numpy / cupy / dask were already covered.

Cat 2 (NaN / Inf / nodata)

• NaN parity for cupy and dask+cupy backends.
• all-NaN raster on every backend.
• +/-Inf pins on every backend. The numpy / dask / dask+cupy backends silently absorb Inf cells into adjacent finite polygons (issue polygonize numpy/dask backends silently absorb +Inf/-Inf cells #2155); cupy emits them correctly. The pins lock the current asymmetric behaviour so the fix is visible as a test diff.

Cat 3 (geometric edge cases)

• 1x1 single-pixel raster on all four backends + geopandas.
• Nx1 single-column raster on all four backends. Exercises the nx==1 padding path at polygonize.py:565 and the cupy nx==1 numpy-fallback at polygonize.py:671. Neither code path had direct test coverage before.
• 1xN single-row raster on all four backends.
• All-equal-value raster on all four backends.

Cat 4 (parameter coverage)

• column_name= non-default value across geopandas / spatialpandas / geojson return types.
• Error paths: bad connectivity, bad transform length, mask shape mismatch, mask underlying-type mismatch.

Cat 5 not applicable: polygonize returns lists / dataframes, not a DataArray with attrs to propagate.

Source-bug surfaced

Filed #2155 for a real bug surfaced by the +/-Inf tests: numpy / dask / dask+cupy backends silently absorb Inf cells. Per sweep rules the source fix is NOT bundled with this test-only PR.

Test plan

• python -m pytest xrspatial/tests/test_polygonize_coverage_2026_05_19.py -> 58 passed.
• python -m pytest xrspatial/tests/test_polygonize.py xrspatial/tests/test_polygonize_coverage_2026_05_19.py -> 172 passed, 13 skipped (pre-existing skips for optional deps).
• CUDA host: cupy and dask+cupy tests executed locally.