Guard QGL2PaintEngineEx/QGLPaintDevice against using unset context

QGL2PaintEngineEx::end() calls endPaint() on the device, which may end up
calling QGLContext::swapBuffers() on the context, a call which we have
documented to require makeCurrent() for any further GL commands:

  Call this to finish a frame of OpenGL rendering, and make sure to
  call makeCurrent() again before issuing any further OpenGL commands,
  for example as part of a new frame.

QOpenGLContext also documents that:

  To be able to support certain platforms, QOpenGLContext requires that
  you call makeCurrent() again before starting rendering a new frame,
  after calling swapBuffers().

As a consequence, we need to guard the GL usage in QGLPaintDevice::endPaint(),
as well as the GL reset in QGL2PaintEngineEx::end().

Change-Id: Ib66e91cce12fbaea32ed3c30141150fa15907187
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>

src/opengl/gl2paintengineex/qpaintengineex_opengl2.cpp

@@ -2099,9 +2099,8 @@ bool QGL2PaintEngineEx::end()
     d->device->endPaint();
 
     ctx->d_ptr->active_engine = 0;
-
+    ctx->makeCurrent();
     d->resetOpenGLContextActiveEngine();
-
     d->resetGLState();
 
     delete d->shaderManager;

src/opengl/qglpaintdevice.cpp

@@ -126,6 +126,7 @@ void QGLPaintDevice::endPaint()
 {
     // Make sure the FBO bound at beginPaint is re-bound again here:
     QGLContext *ctx = context();
+    ctx->makeCurrent();
 
     ctx->d_func()->refreshCurrentFbo();