[x] consider doing prediction without providing settings, by using settings from comp metadata. Maybe even not allow settings as arguments to prediction function. This would simplify example script significantly, but also requires refactoring of all scripts that uses hatyan.
[x] check which settings are predominant with merge_componentgroups >> made stricter so they have to be the same
[x] it might make sense to drop settings from timeseries object >> would complicate things, since we do need attrs like grootheid, so leave as is for now
[x] expose prediction_perperiod and add unittest (since it is not covered, testbank is green while AttributeError: module 'hatyan' has no attribute 'prediction_perperiod').
[x] make prediction_perperiod accessible via prediction instead of separate function, just like analysis.
prediction_perperiodand add unittest (since it is not covered, testbank is green whileAttributeError: module 'hatyan' has no attribute 'prediction_perperiod').prediction_perperiodaccessible viapredictioninstead of separate function, just like analysis.