In old-style Indic OT standards, the post-base Halants are moved after
their base.  Emulate that by moving first post-base Halant to
post-last-consonant.

Brings test-shape-complex failures down from 88 to 54.  Getting there!

Old HarfBuzz test suite always shaped as left-to-right and hence had wrong
0x14db, direction expected glyphstring for N'ko.  Doh!

Failures down from 92 to 88.

Brings test-shape-complex failures down from 104 to 92.  Way to go!

Is the categorization correct?  I don't know.

Number of failing shape-complex tests goes from 125 down to 94.

Next: Add Ra handling and it's fair to say we kinda support Indic :).

Find the base consonant and apply basic Indic features accordingly.
Nothing complete, but does something for now.  Specifically:
no Ra handling right now, and no ZWJ/ZWNJ.

Number of failing shape-complex tests goes from 174 down to 125.

Next: reorder matras.

Buffer var allocation coming into shape

Looks *so* much nicer!

Buffer debugging coming soon.

I've messed up a lot of stuff recently, different parts of the
shaping process are stumbling on eachother's toes because
manually tracking what's in which buffer var is hard.  I'm
going to add some internal API to track those such that mistakes
are discovered as soon as they are introduced.

This frees 16bits in the glyph_info struct during the ot_layout process.
We can use the freed space in the shapers now.

Fallback normalization is complete and working now!

Add tests.

This allows the Indic shaper to request decomposed characters.  This will
handle split matra for free.  Other shapers prefer precomposed
characters.

Lets not worry about performance for now...