To fix bug 6982632 we had to make the test only fail optionally if the rendering methods threw exceptions. While this does not violate any specification or stated
promises about the behavior of these methods, it does represent a change in behavior that is a bit more annoying than the silent failures that used to happen.

We should consider installing fake NOP loops instead of setting the loops to null when the SunGraphics2D is invalidated. This will preserve the lack of exceptions with a more benign failure (no output).