As explained before, Nvidia just grabbed the whole dynamic init and made arrays
of it, without actually optimizing it.
The second part of the streamline aims to fully de-duplicate that.
- Completely remove all already set registers for DC/DISP/WIN.
- Do not touch other windows when a specific window is setup.
- Init Window D also together with A/B/C since code is made for DISPA.
- Add missing increase for syncpt 1.