Combine DiffNets with existing functionality

[martinvoegele]

We would like to run DiffNets from the PENSA package and make it work with existing features.
In addition, the currently implemented methods might benefit from using featurization functions that so far are only implemented with DiffNets. First steps in this direction:

• Move DiffNets preprocessing/featurization to the corresponding modules and adapt the output format.
• Adapt DiffNets input to use PENSA-style features.
• Harmonize functionality across feature readers (e.g. whitening).

[martinvoegele]

The corresponding branch is 20-combine-diffnets.

I have moved the following:

• data_processing functionality to preprocessing.whitening_xyz
• CLI scripts to the scripts folder
• tests to tests/test_diffnets*.py

Some dependencies will still need to be adapted accordingly.

The submodule diffnets.analysis is not yet included in the package because we want to avoid a dependence on enspara if possible.

[martinvoegele]

We found that including this would introduce too many dependencies that would end up making PENSA too inflexible to be used with any kind of other ML frameworks. So I am closing this for now.