Adding Clustering as a preprocessing step for sampling/outlier detection

Leveraging a clustering-based anomaly detection tool could help in a few regards as a preprocessing step.
- Outlier detection can be used to identify outliers within the feature space and they can be subsequently dropped or their values can be clipped.
- As part of our downsampling tools, if there is an excess of data points for the majority class in an imbalanced dataset, this tool could remove some redundant data that is extremely close in feature space to prevent the model from over-learning on them.

This would require design to identify the full scope of this tool and whether it would need to be broken down into multiple implementations.