Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

We should add a bot for it
Part of #1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Part of https://github.com/harfbuzz/harfbuzz/issues/1652

Add a flag to hb_buffer_t to prevent the insertion of dotted circles on incorrect character sequences. 

Current behavior unchanged if this flag is not set (and it isn't by default).

Fixes https://github.com/harfbuzz/harfbuzz/issues/1556

Adjust tests.

Fixes https://github.com/harfbuzz/harfbuzz/issues/1532

The override_features is used to override features that are normally
discretionary features, but in a specific shaper are for various
reasons desired to be bolted on or off, because they've been used
for inherent shaping.  As such, it makes sense that they also
override user features.  Ie. if user turned 'liga' on, we don't
want Khmer shaping to become broken...  Or turn 'clig' off...

Fixes https://github.com/harfbuzz/harfbuzz/issues/1310

Geeza Pro for example, relies on that for fancy mark positioning.

Fixes https://github.com/harfbuzz/harfbuzz/issues/1405

Cross-kerning can only take care of positioning vertically.  It doesn't
adjust mark advance...

Such that for Indic-like scripts (eg. Khmer), we don't do any fallback mark
advance-zeroing / positioning, but we do for Latin, etc.  Reuses preferences
of our script-specific OpenType shapers for those.

Fixes regression: https://github.com/harfbuzz/harfbuzz/issues/1393
Which means, fixes again: https://github.com/harfbuzz/harfbuzz/issues/1264
While not regressing: https://github.com/harfbuzz/harfbuzz/issues/1357

That's what Apple does, and it wouldn't degrade our OpenType performance.

Part of https://github.com/harfbuzz/harfbuzz/issues/1348

Fixes https://github.com/harfbuzz/harfbuzz/issues/1348

Fixes https://github.com/harfbuzz/harfbuzz/issues/1357

Such that we don't accidentally use info not in the cache key.

No other shaper will need shape_plan_data, by definition.  So, remove
abstraction layer and always create hb_ot_shape_plan_t as part of
hb_shape_plan_t.

Finally:  Fixes https://github.com/harfbuzz/harfbuzz/issues/1356

Test case:
$ ./hb-shape GeezaPro.ttc -u U+0628,U+064A,U+064E,U+0651,U+0629
[u0629.final.tehMarbuta=4+713|u064e_u0651.shaddaFatha=1@0,-200+0|u064a.medial.yeh=1+656|u0628.initial.beh=0+656]

The mark positioning (kern table CrossStream kerning) only works if deleted
glyph (as result of ligation) is still in stream and pushed through the
state machine.

Happens even if the cross-stream kerning is for cursive attachment only.  Oh well..

Ouch!