ref(grouping): Switch to using split enhancements

[lobsterkatie]

This switches from using non-split enhancements (and sometimes also calculating split enhancements) to just using the split enhancements. In the time since the experiment was enabled, across over 150 million events, the results of the split enhancements have matched* those of the non-split ones, and the split enhancements have been running on average about 45% faster than the non-split ones.

• To be more precise, they have matched except in cases where the split enhancements have given different/better hints, in ways we expect:

• Because the rules are split, hints about marking things in-app or out of app only include the +/-app part of the rule, and hints about ignoring/unignoring only include the +/-group part.
• In cases where both an in-app and contributes change have happened, both are now included in the hint.

This can be seen in the snapshot changes included in this PR.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #92180      +/-   ##
==========================================
- Coverage   87.90%   87.90%   -0.01%     
==========================================
  Files       10226    10226              
  Lines      585857   585807      -50     
  Branches    22752    22752              
==========================================
- Hits       515016   514952      -64     
- Misses      70412    70426      +14     
  Partials      429      429