v2.0.0

Added

• Added proper __repr__ descriptions to all models

Changed

• Changed the ordering of Dataset.split to X_train, X_test, y_train
  and y_test, to agree with scikit-learn
• Moved some Dataset attributes to the private API