Braille regression in ben-unicode-internal (and some other unicode char groups)

[Stephen J. Turnbull]

Jeff Sparkes writes:

 > there were regressions in various character sets.  Since they
 > displayed in b29, the fonts must be available.

Not necessarily.  It depends on how Ben's workspace translates from
Unicode to glyph indexes, which has almost certainly changed because
of the change from Mule internal to Unicode.

I imagine that the structure of faces and fonts hasn't changed, so you
need to look in objects-{platform}.c and related files to see how
font_instances are constructed, and then in faces.c and possibly
redisplay-{platform}.c to see how fonts are accessed.  (I don't know
whether he did the refactoring before or after the unicode branch, but
those files have been renamed in the mainline.)