Implement alternatives to SMOTE for rebalancing training data

[gilbertocamara]

Describe the requested improvement

The SMOTE method has become a reference for reducing sample imbalance for training samples. More recently, a number of improvements to SMOTE have been proposed in the literature. One of them is SMOTEWB, which uses a boosting method to determine the appropriate number of neighbors for generating new samples with SMOTE. This algorithm is described in the paper "A novel SMOTE-based resampling technique trough noise detection and the boosting procedure".

The authors of SMOTEWB have implemented their algorithm in the R package SMOTEWB. Alternative resampling methods are also available in the package. It is relevant to include them in sits.

Associated sits API function
sits_reduce_imbalance