Fix/ci autotune

[laggui]

[WIP]

Locally f16_ty::ad_broadcast::tests::matmul_broadcast test w/ vulkan completes in 202.91s.

Autotune checks run on every op, even if not selected. Removing the checks drastically reduces this single test time to 1.56s.

Further limiting the max tunables per group (config WIP) can further reduce the time to 1.44s (negligible without the autotune checks). To keep the autotune checks we should only execute on the limited tunables, not all ops.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 35.19%. Comparing base (16be063) to head (9b90d2c).
Report is 1 commits behind head on main.

❌ Your project check has failed because the head coverage (35.19%) is below the target coverage (80.00%). You can increase the head coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3362      +/-   ##
==========================================
- Coverage   35.20%   35.19%   -0.01%     
==========================================
  Files         342      342              
  Lines       53144    53164      +20     
==========================================
+ Hits        18709    18712       +3     
- Misses      34435    34452      +17     

☔ 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.