Speedup Filter[s]Aggregator using competitiveIterator similar to dynamic pruning #19621
Description
Is your feature request related to a problem? Please describe
I noticed that FiltersAggregator:
- first builds the Weight and Bitset of matching documents for each filter
- iterates over docId matching query one-by-one, checking it against each filter
Describe the solution you'd like
We can do much better by efficiently skipping over docIds using competitive iterator of DisiPriorityQueue similar to dynamic pruning, except we don't prune but keep advancing the docId for scorer.
Related component
Search
Describe alternatives you've considered
No response
Additional context
No response