bug in gdb

[Jim Blandy]

> Found a bug in gdb.  This seems to be the correct fix.  (I give the
> code unconditionally to the FSF for what ever they wish to do with
> it..., yadda, yadda, yadda).  It does not seem to warrant much
> fanfare. 

For a change this small, we don't need a copyright assignment.  It's
not large enough to be copyrightable.

> Changelog entry:
>   Fixed a bug where the typevec_size was not being reset to zero after
>   the typevec itself had been reset.

The fix looks reasonable, but can you give me an identifier to
demangle that makes the bug appear?