[ ] Implement plotting utility for generic predictions and run it after inference, so with submitting the script on the cluster, all results can be inspected.
[ ] Implement a default posterior predictions method, which simply plots all observations against their predictions for the different data variables