nanogui: Thread: one more item

Shane,
    One more item.

The reason that fltk seems complicated is that, unlike X, most
windows programs are ONLY supposed to draw during WM_PAINT.
That's it.  See, then the "original" drawing and the "repaint" drawing is
exactly the same.  With X, it's different.

So, with fltk, they screw around with all sorts of region management
to get it to NOT paint during WM_PAINT, since all the painting is done
outside of that.  Get it?  Now, the only difference to microwindows
is calling GetDC rather than BeginPaint to get the hdc.  Of course, the 
WM_PAINT handler has to coax all the fltk routines to repaint again,
but that's an internal C++ objects thing.

Now, fltk uses the CS_OWNDC so that it doesn't have to ask for
a GetDC all the time, because that's like X.  But we always have to 
get a new one, since microwindows computes the clip region at GetDC
time.

Thought you might want to know all this ;-)

Greg