Code refactoring - Core and Scheme

[shelllbw]

Refactor and optimise training, prediction, and evaluation pipeline to improve readability and performance

• Refactor core learn() and predict() classes; merge GNN_train() and GNN_predict() into the train() and predict()
• Move model initialisation to learn(), use deep copies to initialise models for kfold, bootstrap and ensemble schemes
• Refactor and optimise all training schemes
• Refactor and optimise all prediction schemes
• Improved screen output and result formatting
• Default parameters for all training schemes
• Review AEGAN learn scheme, check for its correctness
• Integrate evaluation scheme to the new framework; See if the scheme can be initialise at learn()
• Potentially optimise evaluation codebase