Add class for custom transforms to adapter, similar to LambdaTransform.

[vpratz]

Closes #362.

This PR reintroduces the features that were present in LambdaTransform, but only allowing registered functions. While being stricter, this allows for closer scaffolding and raising errors early on, so that users cannot provide functions that will not be (de)serializable later on.

As there are a few failure modes, the focus is on providing detailed error messages to enable users to solve problems without external help.

Open questions:

• When writing the code, I chose the naming to make it very clear that you cannot pass any functions, but only serializable ones. In my impression, I overdid it a bit, and the point should be sufficiently clear, even when we rename serializable_forward/inverse_fn to forward/inverse again. What do you think?
• I opted for a method in the Adapter, as it helps in easily providing the user with the FilterTransform functionality. Is this ok with you, @LarsKue
• As always, naming suggestions are welcome.

[codecov]

Codecov Report

Attention: Patch coverage is 90.41096% with 7 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...apters/transforms/serializable_custom_transform.py	89.55%	7 Missing ⚠️

Files with missing lines	Coverage Δ
bayesflow/adapters/adapter.py	80.87% <100.00%> (+0.42%)	⬆️
bayesflow/adapters/transforms/__init__.py	100.00% <100.00%> (ø)
...apters/transforms/serializable_custom_transform.py	89.55% <89.55%> (ø)

[LarsKue]

Very awesome changes, thank you for covering this so thoroughly!

I do see the need for some minor changes, particularly in the tests. See individual comments.

[vpratz]

Thanks a lot for the review, I have addressed your comments. The tests are currently failing because I pulled in the changes from dev. If you have no objections, I will merge when the tests on dev are running again.

[LarsKue]

@vpratz Yes, dev is now fixed. Can you please merge dev into this again and if tests pass, merge?