Sort out display() being sometimes called from the EDT inside OpenGL

[processing-bot]

Created by: benfry

Inside PSurfaceJOGL, we've got:

public class DrawListener implements GLEventListener {
    private boolean isInit = false;

    public void display(GLAutoDrawable drawable) {
      if (!isInit) return;

      if (display.getEDTUtil().isCurrentThreadEDT()) {
        // For some unknown reason, a few frames of the animator run on
        // the EDT. For those, we just skip this draw call to avoid badness.
        // See below for explanation of this two line hack.
        pgl.beginRender();
        pgl.endRender(sketch.sketchWindowColor());
        return;
      }

…which is an inauspicious beginning, and it doesn't get much better from there.

We need to figure out:

1. why those calls are coming from the EDT, and
2. how we should handle them properly

I think the answer to the first question is that any time the window manager (OS) does a UI update, rather than the Animator thread doing its thing. That said, we need something definitive on whether we should be drawing there (at the moment, the understanding is that we shouldn't) and how it should be handled instead (i.e. re-queue the event in another thread).

But this is also related to #370 to some degree, where it's unclear to what degree we're paying for badness like this lurking in the code, or something that's not entirely our fault.

[processing-bot]

Created by: benfry

Lots of similar issues in PGraphicsOpenGL.readPixels(), where Exceptions are just ignored, presumably to paper over issues during resize events.