machine learning integration

[FAMILIAR-project]

Researchers of DiverSE have trained machine learning models to predict size or build status.
We should integrate them as part of the website.

There are several things to develop:

• a predictor of size in which the user simply has to drop a .config
• for each configuration of the database, add a new "size predictor" information that would call the predictor (we can also report the prediction errors by comparing the prediction with the actual size)
• for each kernel version, report so-called "feature ranking list" using the machine learning models
• for each kernel version, report so-called decision trees and feature importances about build status