ligatures in 8r encoded fonts

[Lars Hellström]

As I have now had the time to look at it, I can definately say that it is a
genuine bug (I think it's the first code bug reported about v1.9). It is
not the ETX that is wrong (although you can work around the bug by
modifying this), but fontinst itself. The problem is that the v1.9
\vpl_ligature assumes that every glyph that has been assigned a slot really
exist (the v1.8 version checked), whereas \end_assign_slot makes an
assignment regardless of whether the glyph exists or not.

I tend towards changing \end_assign_slot, as I can't see the point with
assigning nonexistent glyphs to slots.

Lars Hellström