Implement order-1 entropy ranking.
Order-n entropy ranking needs all locations from the parsing, not just the count so I had to re-write the architecture.
While I was at it, I redefined how match finding (previously referred to as counting) should work.
Changelog:
Add the Match object to represent matches instead of an array of words.
Rewrote the IRanking and IMatching interfaces to use the new types
Created a unified BWDIndex, containing the SA, LCP and bitvector
LCPMatchFinder automatically detects the max word size
Implement order-1 entropy ranking. Order-n entropy ranking needs all locations from the parsing, not just the count so I had to re-write the architecture. While I was at it, I redefined how match finding (previously referred to as counting) should work.
Changelog: