Keep track of scale separation components
At the moment, during the calibration we don't know what component of the time scale separated timeseries (multi-day mean or daily anomalies) we are dealing with. In some sense, that is not a problem because the configuration determines everything and no decisions are based on that distinction. However, it would be nice to add an attribute to the respective cubes so that we can keep track and exploit that also for things like temporary file names.