Add regressor with option for XGBoost or CatBoost

[HellenNamulinda]

This pull request is an impplementation of a regressor class capable of using either XGBoost or CatBoost with hyperparameter optimization. The class provides methods for training, predicting, evaluating (using metrics such as Mean Squared Error, Root Mean Squared Error, Mean Absolute Error, R-squared Score, and Explained Variance Score), and explaining the model. Changes:

• Implemented a regressor with the option to select either XGBoost or CatBoost as the model algorithm.
• Utilized Optuna for hyperparameter optimization.
• Tested using data from Chembl33 for pf3d7_ic50 of size 2082, after removing duplicates(9f520ecaa7a6bcfea6deffc47742b1ddb1bd32cb)
• Generated sample SHAP plots to visualize the shap values(902c5a95c4dd1b21d07259a02557817f8bc2208e)

[miquelduranfrigola]

This looks pretty good, @HellenNamulinda . I've merged the PR but not tested it.

[HellenNamulinda]

The file format for storing evaluation metrics has been updated to json instead of txt(fe93e503588c0357baa1b644d929441ffff150fd) This will reflect in the next PR.