perf: implement fast path for union many row id tree map

[chebbyChefNEQ]

Use roaring::MultiOp for faster union of bitmaps.

TODO:

• add a benchmark to figure out how much we are saving from this
• add a row id tree map benchmark to compare union_all and regular loop

[chebbyChefNEQ]

this is essentially betting that full is rare. We might want more comprehensive benchmarks to tune the behavior here

[wjones127]

Yeah Full is mostly to represent the deletion prefilter. For bitmap I think your right it should be rare.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 92.68293% with 3 lines in your changes missing coverage. Please review.

Project coverage is 77.32%. Comparing base (ceaf49c) to head (65502fe).

Files with missing lines	Patch %	Lines
rust/lance-core/src/utils/mask.rs	91.42%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3087      +/-   ##
==========================================
+ Coverage   77.31%   77.32%   +0.01%     
==========================================
  Files         240      240              
  Lines       79322    79358      +36     
  Branches    79322    79358      +36     
==========================================
+ Hits        61326    61365      +39     
- Misses      14821    14825       +4     
+ Partials     3175     3168       -7     

Flag	Coverage Δ
unittests	77.32% <92.68%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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