WIP

Merges feature-extractors outputs and reduce dimensionality into a desired output size, to serve as input size for a predictor (or another ML system). Uses plug-ins for the dimensionality reduction technique.