What does this implement/fix? Explain your changes.

Major refactoring and clean up of existing functionality:

• folder reorganisation Folder structure #229
• sklearn-compatibility checks, run sklearn's estimator checks
• consistent input checks for TSC Implement consistent input checks #98
• consistent unit checks for TSC Update unit testing #251
• consistent is-fitted checks for TSC
• inherit from sktime's BaseEstimator
• linting Enforce linting #256
• fix warnings

Comments

I had to remove a few classes which I couldn't make compliant with the basic scikit-learn API without major work, including

• transformers/spectral_based, see Refactoring transformers/spectral_based  #224 for refactoring and adding it back
• removed our version of GridSearchCV as it was essentially the same as the one in scikit-learn
• HomogeneousColumnEnsembleClassifier, but we still have the more flexible ColumnEnsembleClassifier
• removed optional check_input kwarg (or parameter passed to fit)
• moved strategies and tasks from highlevel/ to benchmarking/, removed highlevel/
• removed our Pipeline, but sklearn's pipeline should be fully compatible