RFC: C++ overloadin fixes and a type "char" tweak

[Michael Elizabeth Chastain]

Hey, I've read this code recently!

  http://sources.redhat.com/cgi-bin/gnatsweb.pl?cmd=view&database=gdb&pr=872
  gdb fails overload resolution with TYPE_CODE_INT types

Proofread but not tested.  The gdb.c++/overload.exp part is approved 
no matter what.  Those setup_xfail's are a flat lie, GDB has correct
information and then makes the wrong overload choice on its own.
The gdb part looks fine to me but I don't have resources to test it
for a while.
 
Michael C