current Plug and Play PR - 1 set + preparation for further ELAN cleanups

[mamehaze]

new NOT WORKING machines

VJ Starz Dance Mat [Team Europe, David Haywood]

• fixed a regression with qix in rad_sinv, which suggests a difference in EU3A05 and EU3A13 behavior when banking
• use subclass for ELAN EU3A13 video as there seem to be some subtle differences, removed hacks in driver

[mamehaze]

This is the first stage in refactoring / tidying up the ELAN support.

The software running on these SoCs is quite annoying to work with though, much of it is badly written and assumes default behaviors / states, not writing registers unless they're used etc.

I have experimented with a more extensive clean-up before this, but reverted my progress on that in favour of breaking the work into smaller chunks.