It seems we both are on a different page here, let me explain why:
I believe that if you target aros on only high-end 68k. It will never gain any traction whatsoever. No one besides tinkerers and some few curious will do something with it (like it is now).
Yes, it is easy to write unoptimized code, and rely on speed to cover the holes. You can do that on x86/x64 already. But point is that if you are looking to become a real feaseable replacement on 68k, you need to stop making excuses for features, performance and backwards compatibility. If you want to ignore these, then better drop 68k support alltogether.
If you are looking for high-end stuff that works the way you mention, you already have ArosVision that may suit you, and does its job very well. No need to look any further.
But if you are an average Amiga user, you wont touch aros68k even with a stick. The vast majority of Amigas are worse specced than your proposed RTG & 68020+ hardware.
Not targetting the lower end Amigas (which is the vast majority) is a huge mistake. But that being said, you dont need to provide these low power systems with all the new features aros brings (it should be stupid to believe you can do that). You just need them to be provided with at least what they already have with their aging AmigaOS 3.1 in terms of features and compatibility. This is the stone that I consistenly try to mention, that steps in the way for aros68k to become a viable alternative for most Amiga users. If not ask Amigakit/Vesalia/etc why they still hold in stock, and sell lots of AmigaOS 3.1 floppy sets and dont even dream of replacing them with Aros68k ones (The same can be said of kickstart roms).
The potential is there, but it definately still requires a lot of work and a change of mentality from developers on 68k (if there is still any) to make it happen.