* Use a single QImage copy for actual drawing
* Use std::array and std::unique_ptr for image buffers
* Signal immediately after copying the buffer to internal image
* Only center the cursor when it leaves the emulator screen
* Ignore the very first mouse event that appears after it is captured for the first time after starting
in gleswidget to renderstack (a QStackWidget), which in
turn calls an actual renderer. added ability to target GLES
directly, but this is maybe uneeded.
