[ty] Allow classes to inherit from type[Any] or type[Unknown]

[AlexWaygood]

Summary

This continues our recent work to pay more respect to the gradual guarantee when inferring the MROs of classes. Classes are now permitted to inherit from type[Any] and type[Unknown]: since these are actually intersection types, the same principle applies to these types as was established in #18055.

Test Plan

mdtests added.

I wish I could tell you why the order of the diagnostics in the snapshot is changing as a result of these changes, but -- alas! -- I cannot. It doesn't seem like a massive problem, though.

[github-actions]

mypy_primer results

Changes were detected when running on open source projects

ignite (https://github.com/pytorch/ignite)
+ warning[unused-ignore-comment] ignite/distributed/auto.py:351:75: Unused blanket `type: ignore` directive
- Found 2276 diagnostics
+ Found 2277 diagnostics

operator (https://github.com/canonical/operator)
- error[invalid-base] ops/_private/harness.py:430:26: Invalid class base with type `type[Unknown]` (all bases must be a class, `Any`, `Unknown` or `Todo`)
- Found 203 diagnostics
+ Found 202 diagnostics

mypy (https://github.com/python/mypy)
+ warning[unused-ignore-comment] mypy/stubtest.py:435:35: Unused blanket `type: ignore` directive
- Found 3371 diagnostics
+ Found 3372 diagnostics

dd-trace-py (https://github.com/DataDog/dd-trace-py)
- error[invalid-super-argument] ddtrace/settings/_config.py:768:20: `type[Unknown]` is not a valid class
- error[invalid-super-argument] ddtrace/settings/_config.py:773:20: `type[Unknown]` is not a valid class
- Found 8604 diagnostics
+ Found 8602 diagnostics

scipy (https://github.com/scipy/scipy)
- error[invalid-super-argument] scipy/stats/_continuous_distns.py:68:20: `type[Unknown]` is not a valid class
- error[invalid-super-argument] scipy/stats/_continuous_distns.py:6765:20: `type[Unknown]` is not a valid class
- error[invalid-super-argument] scipy/stats/_continuous_distns.py:8568:20: `type[Unknown]` is not a valid class
- error[invalid-super-argument] scipy/stats/tests/test_distributions.py:295:21: `type[Unknown]` is not a valid class
- Found 7733 diagnostics
+ Found 7729 diagnostics

[AlexWaygood]

That's quite interesting -- a lot of the primer reports are to do with super() rather than inheritance, as our BoundSuper type also users ClassBase as its inner data:

ruff/crates/ty_python_semantic/src/types.rs

Lines 8100 to 8106 in 41fa082

	/// Represent a bound super object like `super(PivotClass, owner)`
	#[salsa::interned(debug)]
	pub struct BoundSuperType<'db> {
	pub pivot_class: ClassBase<'db>,
	#[return_ref]
	pub owner: SuperOwnerKind<'db>,
	}

I think the same arguments about the gradual guarantee apply to super() usage on type[Any] as they do to inheritance from type[Any], though. So this doesn't seem like a problem to me.

(The hits in operator and mypy are both inheritance-related; not all the primer hits are super()-related.)