Fix some Any subclassing in tqdm

[Avasam]

Fixes 4 2 Any subclassing issues by replicating the types used in external modules.
The rich one isn't complete as it was starting to get quite complex (I don't wanna re-implement the entire library's types either!)

Ref: #9491

[AlexWaygood]

The keras and dask changes look great. Mind splitting the rich changes off into a separate PR? They're a little more complex and my brain's going a little fuzzy thinking about nominal vs structural subtyping 😅

[Avasam]

The keras and dask changes look great. Mind splitting the rich changes off into a separate PR? They're a little more complex and my brain's going a little fuzzy thinking about nominal vs structural subtyping 😅

If it feels too complex for our needs, I can always make more things Incomplete. We mostly need the structure just to fix the Any subclassing, detailed types are bonuses.

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉

[AlexWaygood]

Thanks!

[AlexWaygood]

The keras and dask changes look great. Mind splitting the rich changes off into a separate PR? They're a little more complex and my brain's going a little fuzzy thinking about nominal vs structural subtyping 😅

If it feels too complex for our needs, I can always make more things Incomplete. We mostly need the structure just to fix the Any subclassing, detailed types are bonuses.

Maybe we could start off with something simpler for now? We can always refine it in the future :)