The folded text: each character decomposed (NFD), its combining marks dropped and a curly apostrophe made straight,
then the whole lower-cased, so a capital sigma takes its final form by its neighbours, as toLowerCase gives it.
For each UTF-16 unit of text, the half-open interval [start, end) of the source, in UTF-16 units, it came from.
The units one source character yields share its interval (a Hangul syllable yields its three jamo, a character
beyond the Basic Multilingual Plane its two units), and a combining mark dropped after a character widens that
character's interval over it, so a match placed through the map never splits a letter from its marks.
A text folded for matching, with the source interval of each of its UTF-16 units.