Allow PyGraph and PyDiGraph to be annotated as generic classes at runtime

[IvanIsCoding]

Fixes #1345. It should address the core aspects for PyGraph and PyDiGraph. However, custom return types will come in a follow up PR

This makes the following work without requiring from __future__ import annotations:

import rustworkx as rx

graph: rx.PyDiGraph[int, int] = rx.PyDiGraph()

Also, this is worth highlighting: I refrained from adding __class_getitem__ to the type stubs as in the stubs that method is inherited from Generic[_S, _T] I ended up adding the method to the stubs as mypy stub tests disagreed with my reasoning

[coveralls]

Pull Request Test Coverage Report for Build 12770141739

Details

• 24 of 24 (100.0%) changed or added relevant lines in 3 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.005%) to 95.83%

---

Totals
Change from base Build 12770043851:	0.005%
Covered Lines:	18384
Relevant Lines:	19184

---

💛 - Coveralls

[mtreinish]

One quick question about the release note. But otherwise this lgtm it's a pretty straightforward feature and one I could see people using in typing contexts.

[mtreinish]

Does this need to be executed? It seems like a static code-block since we just need syntax highlighting

[IvanIsCoding]

I have no preference either is fine. But I will say the advantage of executing is that it tests the code path.

If you execute that with 0.15.1 you’ll get a runtime error!

[mtreinish]

I'm fine leaving it. We're not super worried about our docs build time right now. If we were though this is the kind of thing we can change.