Add a few more tests for interactions between typing.Protocol and typing_extensions.Protocol

[AlexWaygood]

I was looking at lines like these:

typing_extensions/src/typing_extensions.py

Lines 747 to 749 in af89916

	# Determine if this is a protocol or a concrete subclass.
	if not cls.__dict__.get('_is_protocol', False):
	cls._is_protocol = any(b is Protocol for b in cls.__bases__)

typing_extensions/src/typing_extensions.py

Lines 759 to 761 in af89916

	# ... otherwise prohibit instantiation.
	if cls.__init__ is Protocol.__init__:
	cls.__init__ = _no_init

And realised that I wasn't at all sure how the interaction between typing.Protocol.__init_subclass__ and typing_extensions.Protocol.__init_subclass__ would work out. It looks to me like it's fine, but I thought a few extra tests might be good anyway.

[AlexWaygood]

(There is an edge-case bug there, actually. But I'll investigate/fix it in another PR. If it can be fixed.)

[JelleZijlstra]

Looks good, thanks.

Should I merge or wait for the bugfix you alluded to?

[AlexWaygood]

Merge this -- the bug probably isn't fixable? I think we'll have to document it. Writing up an issue now :/

[AlexWaygood]

Merge this -- the bug probably isn't fixable? I think we'll have to document it. Writing up an issue now :/

#245