Add variant integers to buffer item types.  More cleanup coming.

Not tested at all.

Look forward for next character instead of joining to the last character.

Test case: "مرا" rendered using IranNastaliq.

See email thread "Should MarkToBase attachment zero the mark advance?"
started by Jonathan Kew on 23 August 2010 for details.

Patch by Jonathan Kew.  Mozilla bug #465728.

Kinda hand-wavy right now.  Not tested.

Before, the mask in the buffer was inverted.  That is, a 0 bit meant
feature should be applied and 1 meant not applied, whereas in the
lookups, the logic was positive.

Now both are in sync.  When calling hb_buffer_add_glyph() manually,
the mask should be 1 instead of 0.

hb_font_set_scale() now sets the value to be used to represent a unit
pixel.  For example, if rendering a 10px font with a 26.6 representation,
you would set scale to (10 << 6).  For 10px in 16.16 you would set it to
(10 << 16).  This space should be the same space that the get_glyph_metrics
and get_kerning callbacks work in.

As reported by John Daggett.