Marks enums with values as implicitly final

[sobolevn]

Now enums with values cannot be inherited:

from enum import Enum

class NonEmptyEnum(Enum):
    x = 1

class ErrorEnumWithoutValue(NonEmptyEnum):  # E: Cannot inherit from final class "NonEmptyEnum"
    pass

In runtime it caused this error:

Traceback (most recent call last):
  File "/Users/sobolev/Desktop/mypy/ex.py", line 7, in <module>
    class ErrorEnumWithoutValue(NonEmptyEnum):  # E: Cannot inherit from final class "NonEmptyEnum"
  File "/Users/sobolev/.pyenv/versions/3.9.1/lib/python3.9/enum.py", line 127, in __prepare__
    metacls._check_for_existing_members(cls, bases)
  File "/Users/sobolev/.pyenv/versions/3.9.1/lib/python3.9/enum.py", line 483, in _check_for_existing_members
    raise TypeError("%s: cannot extend enumeration %r" % (class_name, base.__name__))
TypeError: ErrorEnumWithoutValue: cannot extend enumeration 'NonEmptyEnum'

Now runtime and mypy are in sync.

Closes #10857

Part of my "better enum" PR series: #10852

[sobolevn]

I had to refactor this function, it is unrelated, but this is how it was rendered for me:

Vs now:

[sobolevn]

This: #11247 (comment)
Source: https://github.com/trailofbits/manticore/blob/522f8ba31e0c6d7c5d5713ed23a3c48baca7cbf8/manticore/core/smtlib/solver.py#L73-L81

looks like a false positive. Will investigate.

[JukkaL]

Thanks for the PR! I like the detailed tests. Looks good, I only have one question.

[JukkaL]

What about a nested class definition? Would it be fine?

[sobolevn]

Thanks, I've added an explicit test case for it. Any nested class makes enum final.

It mirrors the CPython's behavior:

[JukkaL]

Also there's a merge conflict now.

[sobolevn]

Thanks! Rebased.

[sobolevn]

Thanks, I've added an explicit test case for it. Any nested class makes enum final.

It mirrors the CPython's behavior: