AVX-512 support

[Shnatsel]

Draft because it needs a non-mainline fearless_simd

performance numbers: https://gist.github.com/Shnatsel/9b64215c8f7372bb6362c3650101b0f2

[codecov-commenter]

Codecov Report

❌ Patch coverage is 50.00000% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.97%. Comparing base (e9d8850) to head (ca4fa63).

Files with missing lines	Patch %	Lines
src/algorithms/bravo.rs	50.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #80      +/-   ##
==========================================
- Coverage   99.10%   98.97%   -0.13%     
==========================================
  Files          10       10              
  Lines        1566     1568       +2     
==========================================
  Hits         1552     1552              
- Misses         14       16       +2     

☔ 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.

[Shnatsel]

I looked into the regressions for small sizes. They regress because we fall back to scalar bit reversal since BRAVO operates on LANES * LANES blocks.

I've tried adapting BRAVO to work on smaller sizes, and while I got that to be correct I didn't get any performance benefits from that.

There are two things we could do here:

1. Fall back to a BRAVO implementation with a smaller vector size
2. Optimize the scalar bit reversal

We should probably do both. We already have a menagerie of bit reversal methods from our earlier experiments.