bug#14233: 24.3; Don't constrain frame size to character multiples

[martin rudalics]

> We do have macros like FRAME_MENUBAR_HEIGHT and FRAME_TOOLBAR_HEIGHT
> that can be used.  It is better to have that calculation in one place,
> rather than in each port, so this might be a good time to move it.

If someone wants and knows how to do this cleanly, please go for it.
Currently, these macros are a horrible mess as calc_absolute_offset in
window.c demonstrates.

>> When I maximize a frame, that frame may get a new pixel size which is
>> not necessarily a multiple of the frame's character size.  If I now want
>> to resize that frame's windows (and not leave some spare pixels at the
>> bottom of the frame as we do now) I have to communicate the new pixel
>> size of the frame's root window to the window resizing mechanism.  The
>> function that does that is change_frame_size.
>
> That is one occasion where a pixel-function is needed.  But for most calls, 
pixel precision is not needed.  These are the non-tile/fullscreen/maxmimized cases 
in X and NS.

I need them for the fullscreen/maximized case on X and NS as well when
the screen estate assigned to Emacs is not a multiple of its character
size.

martin