Improvements to similarity-based vector search query

[kaivalnp]

Description

#12679 added a query to perform similarity-based vector searches: matching all vectors scoring above a resultSimilarity to the query vector (in Euclidean space, this can be visualized as all vectors within a radius).

1. Change concurrency model from one-thread-per-slice to one-thread-per-segment
   • See Concurrent rewrite for KnnVectorQuery #12160, Parallelize knn query rewrite across slices rather than segments #12325, Revert "Parallelize knn query rewrite across slices rather than segments #12385 for related discussion on concurrency model of KNN queries, which are also one-thread-per-segment
2. Optimize filtered search by applying conjunction on docs that have vectors
   • Similar to a KNN query
3. More information in query explanation

[kaivalnp]

knnPerfTest.py benchmarks (using mikemccand/luceneutil#542) with Cohere v3, 1M docs, 10K queries:

Baseline with -numSearchThread=2:

recall  latency(ms)  netCPU  avgCpuCount     nDoc  searchType  topK  fanout  traversalSimilarity  resultSimilarity  maxConn  beamWidth  quantized  visited  num_segments  filterStrategy  filterSelectivity  overSample  bp-reorder  indexType
 0.956        0.766   1.529        1.996  1000000      RADIUS   N/A     N/A                 0.78               0.8       64        250     8 bits     2912             5            null                N/A       1.000       false       HNSW

Candidate with -numSearchThread=2:

recall  latency(ms)  netCPU  avgCpuCount     nDoc  searchType  topK  fanout  traversalSimilarity  resultSimilarity  maxConn  beamWidth  quantized  visited  num_segments  filterStrategy  filterSelectivity  overSample  bp-reorder  indexType
 0.956        0.734   1.569        2.139  1000000      RADIUS   N/A     N/A                 0.78               0.8       64        250     8 bits     2912             5            null                N/A       1.000       false       HNSW

Baseline with -numSearchThread=8:

recall  latency(ms)  netCPU  avgCpuCount     nDoc  searchType  topK  fanout  traversalSimilarity  resultSimilarity  maxConn  beamWidth  quantized  visited  num_segments  filterStrategy  filterSelectivity  overSample  bp-reorder  indexType
 0.956        0.757   1.515        2.001  1000000      RADIUS   N/A     N/A                 0.78               0.8       64        250     8 bits     2912             5            null                N/A       1.000       false       HNSW

Candidate with -numSearchThread=8:

recall  latency(ms)  netCPU  avgCpuCount     nDoc  searchType  topK  fanout  traversalSimilarity  resultSimilarity  maxConn  beamWidth  quantized  visited  num_segments  filterStrategy  filterSelectivity  overSample  bp-reorder  indexType
 0.956        0.682   1.595        2.337  1000000      RADIUS   N/A     N/A                 0.78               0.8       64        250     8 bits     2912             5            null                N/A       1.000       false       HNSW