again it looks like not MSX3, but a TurboR replacement would be the actual topic to attack first.
All this talk about advancing the MSX platform, to include OPL4 and V9990 in a next 'standard'... ugh.
The vast majority of MSX users would be happy with a modern, compact, affordable replacement for their old boxes. You know, those big, heavy boxes with ~25 year old IC's inside, failing floppy drives, dried up capacitors in their power supplies, etc, etc.
When I'd take such a box, and -in my head- replace chunks of it with more modern parts (and repeat), then I end up with something like the original 1chipMSX. Which happens to be well designed, well built, and runs almost all MSX software I have thrown at it so far (including some MSX2+ software). Admittedly with a few kinks here & there, but nothing major and getting better all the time.
Basically a 'refresh' of that would be fine - FPGA based, MSX2+ compatible including a fast Z80 (preferably R800-like speed or beyond), upgradeable firmware.
If it has cartridge slot(s), then base unit can be kept 'cheap' and small, and the 5 or 10 OPL4 / V9990 users out there (no offense 
) can plug their cartridges in and enjoy.
Including a discrete VDP in such a new design would feel like a (big!) step back. Better put that effort in improving such VDP's FPGA implementation.
For those users who swear by an 100% accurate VDP or soundchip(s), no problem if they go for a machine a la Gr8bit. But that's not the way forward to get as many MSX users on board as possible. End-user price, actually building the thing, upgrading, fix design flaws, etc, etc, it all gets more expensive / troublesome with a design that's based on many discrete parts. Like those old boxes that have earned their retirement... 
Yes, that is the other thing, the 2+ and TurboR still waiting for software 
In scroll games, the 9958 can make a big difference. Software sprites or YUV backgrounds.
