Dumb question time again. For most Mac apps, how hard would it really be to produce a universal or M1-native binary? I get that fully optimizing to the new platform and taking advantage of all of its features will take time, but unless you've got code that's really working down at the machine instruction set level, isn't this just re-compile, test, and upload to the App Store? (Or however else you choose to distribute.)
Get it done, get it out there, and by actually compiling for the platform instead having to have Rosetta translate, you'll almost certainly be more efficient than if you just stayed x86 and let Rosetta handle it. Boom, you're native, and you can improve over time through normal updates as you optimize for the M1.