This makes `LATIN SMALL LETTER L` confusable with `LATIN SMALL LETTER I`
because `CYRILLIC SMALL LETTER PALOCHKA` "ӏ" (mapping to i) wasn't
confusable with `LATIN SMALL LETTER L` "l" (mapping to l) before.
Also add tests for `str_utf8_comp_confusable`.
See UTS#39 "Unicode Security Mechanisms":
http://www.unicode.org/reports/tr39/
This means that characters with accents or other things around them are
now considered the confusable with the base character.
Fixes#557. Fixes#575.