More efficient counter for query hits

[alexklibisz]

The custom MatchHashesAndScoreQuery currently uses a counter that requires allocating an array of shorts with one entry for every document in the segment to track the number of matches for each doc, and then iterating over that array twice to get the top k docs. This is actually substantially faster than using any sort of hashmap I've found, including primitive hash maps. I've tried hppc, hppcrt, and fastlib, and all of them are least 2x as slow (e.g. a segment with 1.1M docs gets 40 q/s with arrays, 20 q/s with hashmaps). I figure this kind of array setup won't scale forever, but I don't want to change it until there's some comparably fast alternative.

[alexklibisz]

Another library maybe worth trying: https://github.com/real-logic/agrona/blob/master/agrona/src/main/java/org/agrona/collections/Int2IntCounterMap.java

[JanecekPetr]

By far the fastest primitive hashmap implementation I found is https://github.com/leventov/Koloboke. It only contains maps and sets, and the original website is down, but the code quality is still there.

[alexklibisz]

Closing as it falls under #160