Cover write_geotiff_gpu compression modes (zstd default + jpeg) (#1646) (#1647)

* Test write_geotiff_gpu compression modes (#1646)

Cover the documented compression= modes that had no targeted
round-trip tests:

- zstd (the default, "fastest on GPU"): pixel-exact round-trip on
  int32 plus default-codec pinning via TIFF tag 259.
- jpeg (nvJPEG with Pillow fallback): round-trip for 3-band uint8 RGB
  and single-band uint8 with mean-abs-diff bounds; pin compression
  tag 7. Exercises the live nvJPEG encoder on a GPU host, not just
  the Pillow fallback.
- deflate + none: plain round-trips outside the COG / nodata-sentinel
  paths so a regression in the basic tiled assembly is visible.
- Cross-codec parity: zstd, deflate, none must produce pixel-identical
  read-backs for the same input (catches predictor / codec mis-wiring).

11 tests, all passing on the GPU host.

Update .claude/sweep-test-coverage-state.csv to record pass 7.

* Address Copilot review feedback on #1647

- Rewrite test_write_geotiff_gpu_zstd_default_matches_explicit so the
  docstring matches what it asserts: pin the compression tag and the
  decoded-array equality, not byte-for-byte file equality (the writer
  may legitimately vary tile padding/ordering between runs).
- Swap the JPEG RGB test input from random uint8 noise to a deterministic
  smooth gradient (mirroring test_jpeg.py::_gradient_rgb). Tighten the
  mean-abs-diff bound from 50 to 8 for RGB and to 5 for the monochrome
  variant; the looser bound only existed because random noise is the
  worst case for JPEG.
- Add test_write_geotiff_gpu_jpeg_uses_nvjpeg_when_available: spy on
  _gpu_decode._nvjpeg_batch_encode via monkeypatch and assert it fires
  at least once when libnvjpeg is loadable. Without this spy a
  regression breaking nvJPEG would silently fall through to the Pillow
  fallback and the round-trip tests would still pass. The new test is
  guarded by _nvjpeg_only so it only runs on hosts where libnvjpeg is
  actually loadable.

Author: brendancol

.claude/sweep-test-coverage-state.csv

@@ -1,3 +1,3 @@
 module,last_inspected,issue,severity_max,categories_found,notes
-geotiff,2026-05-11,,HIGH,2;3;4,"Pass 6 (2026-05-11): added test_overview_resampling_min_max_median_2026_05_11.py covering Cat 4 HIGH parameter-coverage gap on overview_resampling=min/max/median. CPU end-to-end paths were already covered by test_cog_overview_nodata_1613::test_cpu_cog_overview_aggregations_ignore_sentinel; the GPU end-to-end paths and the direct CPU+GPU block-reducer branches had no targeted tests, so a regression on those code paths would ship undetected. 26 tests, all passing on GPU host: block-reducer unit tests (finite + partial-NaN), end-to-end COG writes for both to_geotiff and write_geotiff_gpu, CPU/GPU parity for to_geotiff(gpu=True), CPU nodata-sentinel regression check, and ValueError error-path tests for unknown method names on both backends. Pass 5 (2026-05-11): added test_degenerate_shapes_backends_2026_05_11.py covering Cat 3 HIGH geometric gaps (1x1 / 1xN / Nx1 reads on dask+numpy, GPU, dask+cupy backends; 1x1 / 1xN / Nx1 writes through write_geotiff_gpu) and Cat 2 MEDIUM NaN/Inf gaps (all-NaN read on GPU + dask+cupy, Inf / -Inf reads on all non-eager backends, NaN sentinel mask on dask read path including sentinel block split across chunk boundary). 23 tests, all passing on GPU host. Prior passes still hold: pass 4 (r4) closed read_geotiff_gpu/dask name= + max_pixels= kwargs (Cat 4), pass 3 (r3) closed read_vrt GPU/dask+GPU backend dispatch (Cat 1) and dtype/name kwargs (Cat 4)."
+geotiff,2026-05-11,,HIGH,2;3;4,"Pass 7 (2026-05-11): added test_gpu_writer_compression_modes_2026_05_11.py closing Cat 4 HIGH gap on write_geotiff_gpu compression= modes. The writer documents zstd (default, fastest GPU), deflate, jpeg, and none, but only deflate + none had round-trip tests; the default zstd and the jpeg (nvJPEG/Pillow) paths shipped without targeted coverage. 11 new tests, all passing on GPU host: zstd round-trip + default-codec pinning, jpeg round-trip on 3-band RGB uint8 + 1-band greyscale, TIFF compression-tag header check across none/deflate/zstd/jpeg, plain deflate + none round-trips outside the COG/sentinel paths, and a cross-codec lossless parity check (zstd/deflate/none agree pixel-exact). nvJPEG path was exercised live, not just the Pillow fallback. Pass 6 (2026-05-11): added test_overview_resampling_min_max_median_2026_05_11.py covering Cat 4 HIGH parameter-coverage gap on overview_resampling=min/max/median. CPU end-to-end paths were already covered by test_cog_overview_nodata_1613::test_cpu_cog_overview_aggregations_ignore_sentinel; the GPU end-to-end paths and the direct CPU+GPU block-reducer branches had no targeted tests, so a regression on those code paths would ship undetected. 26 tests, all passing on GPU host: block-reducer unit tests (finite + partial-NaN), end-to-end COG writes for both to_geotiff and write_geotiff_gpu, CPU/GPU parity for to_geotiff(gpu=True), CPU nodata-sentinel regression check, and ValueError error-path tests for unknown method names on both backends. Pass 5 (2026-05-11): added test_degenerate_shapes_backends_2026_05_11.py covering Cat 3 HIGH geometric gaps (1x1 / 1xN / Nx1 reads on dask+numpy, GPU, dask+cupy backends; 1x1 / 1xN / Nx1 writes through write_geotiff_gpu) and Cat 2 MEDIUM NaN/Inf gaps (all-NaN read on GPU + dask+cupy, Inf / -Inf reads on all non-eager backends, NaN sentinel mask on dask read path including sentinel block split across chunk boundary). 23 tests, all passing on GPU host. Prior passes still hold: pass 4 (r4) closed read_geotiff_gpu/dask name= + max_pixels= kwargs (Cat 4), pass 3 (r3) closed read_vrt GPU/dask+GPU backend dispatch (Cat 1) and dtype/name kwargs (Cat 4)."
 reproject,2026-05-10,,HIGH,1;4;5,"Added 39 tests: LiteCRS direct coverage, itrf_transform behaviour/roundtrip/array, itrf_frames, geoid_height numerical correctness + raster happy-path, vertical helpers (ellipsoidal<->orthometric/depth), reproject() lat/lon and latitude/longitude dim propagation. Note: _merge_arrays_cupy is imported but unused (no cupy merge dispatch in merge()); flagged as feature gap not test gap."