Right now, if some of the anomalies specified in the anomalies argument of the dataset generator generate() call could not be inserted, there is no immediate notification and this leads to unexpected behaviour.
Example: asking for 6 different anomalies, but to generate 5 series with max one anomaly per series.
This should be a warning, e.g. logger.warning('Could not insert all requested anomalies'), or maybe a ValueError in its stronger version.