TO DO: Enrichment vs. pre-calculated background

[lhm30]

Future work should aim to provide enrichment calculations for a pre-calculated background of compounds similar to the previous version, i.e. https://github.com/lhm30/PIDGINv2/blob/master/predict_enriched.py.

This is not as trivial to implement as the previous version, since the combination of both the probability of activity threshold (-p) and the applicability domain threshold (--ad) affect which predictions are made - c.f. the previous version just had the probability of activity cut-off.