extra: tighten canonical for loop handling

extra: tighten canonical for loop handling

I'm not totally sure about this patch.  It's technically more correct and
it does solve a real bug and probably most for loops have a distinct end
instead using a variable...

The problem was this:

	for (i = 0; i < variable; i++) {

But it was saying the "implied max" of variable was s32max which is sort of
bogus and unfortunate.  But anyway, the for loop didn't have a break so
that means it exits with i == s32max.  If we just say that we don't know
how it ends that's more accurate.

Another thing is that these days we have smatch_comparison.c so maybe not
knowing is fine?

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>