> But why would we need that?  Most everything we need to know about
 > display is already tracked by the display iterator, so available even
 > without generating glyphs, and that's what the move_it_* functions do.
 > These function do their job by traversing only small portions of the
 > buffer, just large enough for the job at hand to be done.

I rewrote `fit-window-to-buffer' and `fit-frame-to-buffer' using the
display iterator.  Please have a look at the attached patch.

Thanks, martin