[nclack]

Verify CPU and GPU Zarr output with TensorStore alongside zarr-python. Both single-array and NGFF readback suites use the same 10 codec configurations across three buffered-writer modes. Require c-blosc in every build, remove the unavailable-codec stub and conditional skips, and expose the dependency through the installed CMake and pkg-config metadata.

A shared three-level NGFF fixture checks every advertised level against an analytic reference, validates axes and physical scales, and covers complete and partial shards. Missing shards, corrupt payloads/indexes, and incorrect scales must fail; renamed levels must remain discoverable through NGFF metadata. CI installs the reader dependencies before CTest. The fixture and validation helpers can be reused for #34.

Validation:

• Seven focused CPU checks and four GPU checks passed (CUDA 13.2, RTX 5070). Each backend covers 30 codec/buffering combinations in both readback suites.
• Missing c-blosc fails configuration; an installed-package consumer compiled and ran with CHUCKY_BUILD_SHARED=ON, and rejected a disabled Blosc dependency.
• C formatting, Python syntax, and whitespace checks passed.

Closes #35.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
see 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.