Facet and Stat Refactor

[mdavis95]

Closes #94

• Improve facet and facet stat performance

  • Some tests show 30-40x improvement in large indexes with many facets

• Store facets in new binary doc values field that is allows high performance when there are many facet fields in the index that are not used in every query

  • Fall back to old facet handling in existing indexes (but at a high performance penalty)
  • Format is [dimOrdinal dimOrdinalLength ordinalVal0 .. ordinalValN]* that allow facets dimensions that are not requested ordinalValues to be skipped without converting them from bytes to integers

• Unify count request, stat, and stat facet logic into an aggregation handler

  • OOP cleanup of facet and stat logic
  • change facet indexing and searching to not use the lucene facet config helper class

• Use koloboke hash maps for high performance faceting

• bump cache size to 32 million for facet indexing (in lucene 9.4 it was unlimited, defaults to 4k in lucene 9.5)

• Avoid unnecessarily conversion of Mongo document to and from bytes using a document container class

• Optimize ResultHelper by removing string split and replacing with a indexOf/substring combo

  • Add more test cases for Result Helper

• bump Junit version from 5.5.2 to 5.9.2

• default zuliarestore to skip associated files and give option to skip, skip existing, or overwrite

• remove mongo database between test runs to ensure test stability

• make zuliadump use createDirectories instead of createDirectory to create recursively