[jasperblues]

Port of liberation-data/drivine4j#633. The analyzer option is drivine-python#442.

Docs only. Beside the paragraph that says loadMatching scores by BM25 as Lucene and Neo4j's full-text index do, a note now says that a dotted name such as org.example.models behaves differently on the two kinds of store:

• On the SQL stores it is words, split at each dot, so example models finds it.
• Neo4j's default analyzer (UAX#29) keeps it as one token, so only the whole name finds it.

Hyphenated names are left out: UAX#29 splits at a hyphen too, so both kinds of store split them alike. In drivine4j the note points at FullTextIndexSpec's analyzer. Drivine for TypeScript and drivine-python don't take an analyzer yet, and their notes say so.