Use abvm/blwm on USE scripts and Khmer

[simoncozens]

• Add constants for list of Indic/USE scripts
• Add USE and Khmer to abvm/blwm scripts

Fixes #566, #565, #522

[anthrotype]

LGTM!

[behdad]

That's because we send it to Indic. Uncomment.

[behdad]

Make sure you handle Myanmar as well.

[simoncozens]

MS suggests using mark for Myanmar, not abvm/blwm.

[behdad]

MS suggests using mark for Myanmar, not abvm/blwm.

Right.

[khaledhosny]

With this PR, my kashida mark to base is moved from mark feature to abvm/blwm and they don’t get applied for Arabic except in HarfBuzz.

[khaledhosny]

I guess this is because unicodedata.script_extension() returns {'Phlp', 'Arab', 'Mand', 'Ougr', 'Syrc', 'Mani', 'Sogd', 'Adlm', 'Rohg'} for 0x0640, but:

1. For characters that can b e used in multiple scripts, it should be added to all relevant features for these scripts, not only one of them.
2. If the font does not have any characters of a given script other than shared ones, it should not be considered (alternatively font’s languagesystems should to be consulted).