Hello,
thought I need that, turned out it isn't exactly needed. "zcurses end;
zcurses refresh;" accomplishes the same goal. More, 5 system
configurations confirmed that endwin() before resize_term(); is needed,
otherwise moves start to be inaccurate (randomly ignore x position). I
think it's a bug in ncurses. Tested OS X ncurses 5.4, Homebrew ncursesw
6.0-2, Arch Linux ncursesw 6.0, Ubuntu 14.04 ncurses 5.9, FreeBSD
ncursesw.so.8.

So why a new sub-command? For versatility. Some edge situations, future
ncurses versions, PDCurses, etc. The sub-command will secure edge cases
and new situations.

If it has "endwin" third param, it does the needed endwin() and refresh.
If "nosave" is used, curses terminal state will not be re-read into
internal variable. I can imagine some edge case for that.
"endwin_nosave" can be used for both effects.

To test resize_term extension availability, 0 for height and width can
be given, $? == 0 or $? == 2 are possible results.

-- 
  Sebastian Gniazdowski
  psprint3@fastmail.com