Make Saturn CD Block implementations to be user selectable

[angelosa]

We currently have three implementations:

• OG HLE and a 2015 failed attempt, which eventually needs to be merged and overhauled;
• An LLE version, which currently do nothing namely lacking serial support in SH1 core;

We can go two routes here:

1. make a listslot option for cdblock component, allowing the user to select between HLE and LLE cores;
2. make a "playground" driver for the LLE version, disassemble it by abusing of the MAME debug internals and generic terminal usage;

[rb6502]

I think ultimately we should pursue the LLE implementation as the final fix. Yabause has working LLE support that we can look at for guidance on the memory map and SH-1 I/O. Going to slot-selectable CD blocks to allow us to debug the LLE (and for users on lower-end systems) is fine though.

[angelosa]

Something that we are aware and that needs a big refactor in the SH1/2 internals is that CD block can transfer data via the various DMAs scattered thru the system which needs a proper DRQ signal.
chaossd does that after "produced by Sega" screen: transfers thru the SH2 internal DMA, checks whenever the size register becomes zero and ultimately fails because it gets wrong data instead of the CD RAM buffers (which we know they are picky being limited in size). More games probably fails due of similar issues with SCU various DMAs (Whizz splash screen logo hang and Sega Ages erratically looping sounds comes to mind)

[CRTRevivalDreamer]

I think ultimately we should pursue the LLE implementation as the final fix. Yabause has working LLE support that we can look at for guidance on the memory map and SH-1 I/O. Going to slot-selectable CD blocks to allow us to debug the LLE (and for users on lower-end systems) is fine though.

For preservation's sake, please dump the CDB104 firmware!