Feat/custom resampler

[wreise]

Reference issues/PRs
None

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Description
Currently, the minus i-th entry in SlidingWindow().resample(y) is the last element from the minus i-th full window.
This PR offers a generalization, where SlidingWindow.target_resampler is applied to SlidingWindow().transform(y)[i] to obtain SlidingWindow().resample(y)[i].
The default SlidingWindow.target_resampler=lambda x: x[-1] makes the default behavior unchanged.

Screenshots (if appropriate)

Any other comments?
I am not sure if this is the right place to introduce that. Maybe another transformer, which inherits from Sliding window would be better?

Checklist

• I have read the guidelines for contributing.
• My code follows the code style of this project. I used flake8 to check my Python changes.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have added tests to cover my changes.
• All new and existing tests passed. I used pytest to check this on Python tests.

[gtauzin]

I made a technical comment, but I am not sure this is the proper way of doing it. It's quite unsafe as depending of target_resampler, the resampling may not be consistent with the way the input is transformed in transform.

[gtauzin]

Parameters logic should not be in the init of a sklearn-estimator. Move the logic in fit and use it to define an attribute as we do for metric_params in diagrams.PairwiseDistances for example