bfd/elf.c: more wrap/overlap/overflow weirdness

[DJ Delorie]

> I think things have gone haywire before you get to this code.

Well, yeah - in real cases, this type of wrap is *bad*.

> I could argue the bss lma has changed, from 0x10000071b to 0x71b.  ;-)

True, but we're using 32-bit vma's, and the message we print is what
counts :-)

Anyway, the original test was for an AT() that caused an overlap.
This is a wrap that messes up our checks.  Different but related
problems?  If so, different warnings.  Either way, the more accurately
we can advise the user, the better off we'll be (and they'll be).
Telling them the lma changed from 0x10000071b to 0x71b doesn't help
them fix the root cause, which leads back to the original problem - is
wrapping really a bug?  Or just a strange but legitimate scenario?

> I don't know.  I do know that you'd want to make sure the warning
> didn't trigger on a section right at the end of memory.

Doh!  Fixed.