Difference due to different script data in icu4x vs fonttools.unicodedata

[cmyr]

Tracking down a diff in markkern (for GaMaamli.glyphs) where we are putting a particular glyph (u+02BC, "apostrophemod"/ "MODIFIER LETTER APOSTROPHE") into the default kerning lookup, where fonttools is only putting it into latn (for a font with support for latn&grek.)

fonttools is taking the data from version 16.0.0 of ScriptExtensions.txt but the currently released version of icu4x seems to have used a 15.x version of this file (this is a guess, I've opened unicode-org/icu4x#6034 to inquire.)

I'm going to look into whether or not I can figure out how to manually update this data, which would solve the problem. Otherwise we could either move to have our own implementation of this logic, or else complain upstream and try to get it updated?