Handle empty bodies safely

[ilevkivskyi]

Fixes #2350
Fixes #8005

The implementation is mostly simple, except overloads are a bit tricky and there is still little unsafety can slip through (we can fix it later with another function flag if it will look like worth it, see TODO item in a test). The actual diff is quite small but there are tons of tests, because there are many possible combinations and corner cases, and I wanted to test as much of them as possible.

I am following the logic I previously proposed in #2350. I copy it here FTR:

• Properly check empty bodies as a default
• Add an --allow-empty-bodies flag, we need it for tests (hundreds would fail otherwise, also in other projects like sqlalchemy-stubs and django-stubs)
• Allow empty bodies in stub files (obviously)
• Allow empty bodies if method is decorated with @abstractmethod
• In protocols make functions with empty bodies implicitly abstract (unless this is a stub)
• Add a flag for empty-bodied abstract methods to prohibit calling them via super() (unless they come from a stub)

Note that the stub exceptions in two last points can be summarized as: since we don't know the actual body for something that comes from a stub file, we choose to not give a false positive and stay silent.

Another note is that after #8102 is merged we can turn the hidden flag into an error code, but TBH I am not sure it is necessary. It looks like the flag is more handy, and it makes sense (and feels natural) to use the existing error code [return] (missing return statement) for the unsafe empty bodies.

I am going to try this soon against internal codebases to see if there are some problematic corner cases missing and make any adjustment + tests if needed. Note that mypy/visitor.py has an existing unsafety that this new logic uncovers (there is an existing TODO about this). This requires turning warn_no_return flag off for this file.

Finally, there were couple tiny bugs discovered by tests I added, so I just fixed them in this same PR.

[ilevkivskyi]

Actually I decided not to do this. I will not have time, I will leave the branch if someone wants to work on this.

[TH3CHARLie]

I am going to try to work on this if no one else is currently doing the same.