Support k-fold cross-validation

Success when

The user can set either val_percentage - which will lead to the division of the dataset to training and validation fold or supply k which will lead to k-fold cross-validation. In the latter option mean, threshold, and standard deviation of the ensemble will be returned as the final output of the model