Tag decomposed matras about opentype-shaping-documents HOT 4 CLOSED

The mark-positioning subclasses use left|right|top|bottom (or combinations), which map to pre|post|above|below -- for everybody.

Then, in each script's "shaping characteristics" there are sort-order positions for each of those. But those may not be simple.... For instance, all the Indic scripts say that left/pre matras get sorted into the POS_PREBASE_MATRA position. But Kannada has some right/post matras that get sorted into one position and some that get sorted into another: https://github.com/n8willis/opentype-shaping-documents/blob/master/opentype-shaping-kannada.md#the-knd2-shaping-model

I take it that we should be more explicit about this? Perhaps a table in the "shaping characteristics" subsection?

from opentype-shaping-documents.

I added an expanded note in 905c09d that hopefully clarifies.

from opentype-shaping-documents.

Oh of course, the shaping characteristics! Sorry, failed to connect the dots on that one :)

One more request though, would it be possible to list exactly which scripts require special treatment, particularly if it's just one or two? This adds a certainty and saves the reader from having to carefully scan all ten documents in case they missed something small but vital.

from opentype-shaping-documents.

Good point; it is just Kannada and Telugu AFAICT. Added note in 86eb4d4.

from opentype-shaping-documents.