Fix Zstd decompression negative size issue on 32-bit platforms (Fixes #781)

[mkitti]

This pull request updates the decompress function in numcodecs/zstd.pyx to properly handle frame content sizes that exceed size_t on 32-bit systems. It:

• Stores the result of ZSTD_getFrameContentSize in an unsigned long long.
• Checks for ZSTD_CONTENTSIZE_ERROR, ZSTD_CONTENTSIZE_UNKNOWN, and ensures the value does not exceed SIZE_MAX before casting to size_t.
• Only allocates the output buffer after passing all validations.

This prevents negative sizes from being passed to PyBytes_FromStringAndSize, resolving system errors on 32-bit platforms.

Fixes #781

Assistance provided by GitHub Copilot Chat, an AI tool by GitHub, August 2025.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (f42a233) to head (6ab3b78).

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #782   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           64        64           
  Lines         2804      2804           
=========================================
  Hits          2804      2804           

🚀 New features to boost your workflow:

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

[mkitti]

https://github.com/zarr-developers/numcodecs/actions/runs/17060350247/job/48365901555?pr=785 confirms that this fix addresses #781