[COMMIT] Use char-tables, not vectors, to instantiate the display table specifiers

[Aidan Kehoe]

I’ve committed this patch, with a couple more small corrections. See
http://hg.debian.org/hg/xemacs/xemacs?cs=1f0aa40cafe0 and
http://hg.debian.org/hg/xemacs/xemacs?cs=e214ff9f9507 . 

Mike Kupfer followed up on my mail to make the packages I don’t have access
to, work with this--
http://mid.gmane.org/18288.3344.67997.394271@parhasard.net --so this won’t
cause runtime problems with Gnus. Glynn Clements and Uwe Brauer didn’t, so
this change will provoke problems with their packages. I encourage anyone
with access to those packages to make those changes. 

 Ar an ceathrú lá is fiche de mí na Nollaig, scríobh Aidan Kehoe: 

 > This change has the advantage that we can do helpful stuff with Unicode
 > error octets, like displaying them as, say, the corresponding Latin 1 or
 > koi8-r character (depending on language environment) in red. More generally,
 > without this patch, display tables couldn’t handle anything non-Latin-1,
 > which made them of limited usefulness.

-- 
¿Dónde estará ahora mi sobrino Yoghurtu Nghé, que tuvo que huir
precipitadamente de la aldea por culpa de la escasez de rinocerontes?