More on CFF CIDFonts

[Tom Kacvinsky]

I kept getting errors of the sort "invalid local subr number."  for
some of the CID keyed fonts I have.  Moreover, these particular fonts
are prone to having glyph overlap.

I finally traced this down to the fact that in a CID keyed font, one
can have multiple private dictionaries, and each of those private
dictionaries then has its own local subr offset.  I guess (I haven't
delved that deeply into it) that the FDSelect and FDArray entries are
not being parsed, and hence, we aren't accessing the local subrs
needed by these glyphs.  Which makes it easy to see why the metrics
might be off!

More on this later, as I get tings sorted out -- I have another CFF
disassembler (not robust -- won't recursively decode subrs) which
makes easy to see what is going on in a font.

Tom