It’s quite possible that Apple does clock-rate increases in between new architectures, too. That’s how we used to do things. You take an existing design, fix bugs in it, change a few things that are low hanging fruit but which you didn’t have time to do before tape out, and hammer on critical paths to buy yourself a bit more product at the high end of the Schmoo plot. Heck, for all we know Apple is stockpiling die that test in the highest bins so that they can use them as a spec bump going forward
I also see it as very much possible going forward. M-series are still a bit raw around the edges and the mobile-only heritage is clearly seen in M1. Who is to say that future Apple Silicon versions won't be more adaptable to different computing environments?