Improve tutorial on cross-validation and anomaly detection

[mcontisc]

The notebook on cross-validation and anomaly detection using the ACD model (https://mpi-is.github.io/probinet/tutorials/ACD.html) can be improved. Some ideas are for instance:

• use a smaller network so that the cross-validation routine is faster and the visualizations clearer
• analyse a real-world dataset with injected spurious edges and test their recovery
• represent the network with the inferred node communities and anoumalous edges