Add means for evaluation/calibration of embedding models

We need to have means for reserving a validation set (or even make a k-fold cross validation) for the embedding models (usually based on link prediction, i.e. reserve sets of links for validation).

Useful references:

- https://stellargraph.readthedocs.io/en/stable/demos/calibration/calibration-link-prediction.html
- https://neo4j.com/docs/graph-data-science/current/algorithms/ml-models/node-classification/