Dynamic threadgroup memory

[christiangnrd]

Surprisingly it's somewhat functional. Only macOS 15+ since the global dynamic threadgroup memory is not available before then. This isn't the only way to get this dynamic threadgroup memory, but I tried this approach as my first attempt since it seemed the most similar to how static threadgroup memory is implemented.

The Metal interface takes an Integer or a Tuple of the size of the allocation, which is then aligned to the next multiple of 16.

Kernels silently fail under shader validation (which are caught in the tests since the output doesn't match expected results.

Close #701

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.35%. Comparing base (356e7d2) to head (a76f582).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #750      +/-   ##
==========================================
- Coverage   81.74%   81.35%   -0.40%     
==========================================
  Files          66       63       -3     
  Lines        3145     3138       -7     
==========================================
- Hits         2571     2553      -18     
- Misses        574      585      +11     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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