Anomaly detection is running

[JonasFrey96]

Tested anomaly detection:

• To use anomaly detection one has to simply change the model within the cfg
• I added an additional parameter within the confidence generator, I will explain to you in person the following drawing:

Issues:

• within the default.yaml we have a parameter that is used to overwrite the confidence generator used for anomaly_detection and the normal one in our confidence - this is fine but may also be confusing given that you may want to parameterize those two things separately.
• check with robin given that the learning performance is sub-ideal for the flow model, which is independent of the confidence_generator / scaling.
• for the confidence_generator, we are currently running the latest_measurment, but I would assume using the last 5 measurements or some window may be better. I think this should be implemented, but I did not test it.