[PATCH]: realloc insn table in df_modify_insn, if necessary

[law@redhat.com]

  In message < 878zg52x10.fsf@cgsoftware.com >you write:
  > Call df_insn_modify on a new insn.
Right.  But is that considered a valid thing to do?

  > Just as you've worked more with GCSE than I have, i've worked more on
  > the df analyzer than you have.
Certainly.  But something that is obvious ought to be obvious to more than
just you :-)

jeff