I missed that when I skimmed the PEP – it’s mentioned in a single paragraph hidden behind a long argument trying to show that these constructs are confusing. That’s also what the OP (a PEP co-author) used to open the discussion here, and it’s been the entire focus of the discussion so far.
Anyway, when I first learned how to implement a language, some things were drilled into me: features should be “orthogonal”, so the user can combine them in any way they like; and implementation effort is not a reason to avoid a feature properly.
I understand there are different ways to think about such things, but at this point neither of the lines of argument brought up in the PEP has convinced me. If MicroPython thinks this is not worth their effort they are welcome to deviate from Python 3.8 in this respect, but I don’t think a PEP that officially takes this out is a good idea. (Basically the only reason for submitting the PEP now is so they can continue to say “we fully implement Python 3.x.”)