[HarfBuzz] shan ai sequence reordering for Burmese script

[Martin Hosken]

Dear Behdad,

I'd like to report a recent resolution regarding the Shan language using the Burmese script. The basic idea is that for the sequence Consonant + U+1062 + U+1086, it should be rendered with the U+1086 diacritic over the Consonant. The sequence Consonant + U+1086 U+1062 is illegal and should probably insert a dotted circle somewhere. I recommend that the Myanmar shaper reorder the sequence U+1062 U+1086. Here is a test string:

ငၢႆ

This is consistent with UTN#11v4 but would help from being made more explicit.

Yours,
Martin