Segmenter class (exposing strided-roll creation)

[Jeroen Van Der Donckt]

Create a Segmenter class, this class exposes (in a clean interface) the creation of strolls (from the StridedRolling class).

Such a class enables users to create efficiently strided rolls (strolls), opening the door towards integration with;

• deep learning packages (e.g., tsai, 1D or multi-dimensional convolutions)
• raw time-series packages (e.g., tslearn, sktime)

[Jeroen Van Der Donckt]

I think this should go even further (near full decoupling of Segmentation and Feature Extraction functionality)

[Jonas Van Der Donckt]

Also:

• add examples on how to use tsflex for already segmented data