Add array type handling for normalization

[eroell]

Description of changes

Implements different array types for normalization functions which serve as example and guidance on how we implement this for all other functions.

Technical details

Uses singledispatch. Inspired from e.g. scanpy's scale; but less performance engineering, rather focusing on leveraging frameworks such as dask-ml.

[eroell]

Very happy about your input on any of the questions I raised in the comments @flying-sheep :)

[eroell]

Also interested what @nicolassidoux thinks here! :)

[Zethson]

1. You added lots of NotImplementedError messages. I wonder whether we can DRY this more easily?
2. Those error messages should also always suggest which types are support for that function as a sort of solution.
3. adata_to_norm_casted = adata_to_norm.copy() why do we always need those copies? Doesn't the fixture already create new copies?

[eroell]

You added lots of NotImplementedError messages. I wonder whether we can DRY this more easily?

Yess

Those error messages should also always suggest which types are support for that function as a sort of solution.

Good point, the dynamic registries help here again

adata_to_norm_casted = adata_to_norm.copy() why do we always need those copies? Doesn't the fixture already create new copies?

because I just not paying attention to this yet, I'll remove :)

[Zethson]

I addressed the feedback raised in the discussions. Hope I didn't miss any. I'll merge this now but if I forgot something @eroell please make me aware. Thank you!