Add a check for ai.status before deleting datasets

[jay-m-dev]

Checking if the recommender is enabled, and not deleting datasets if it is, since this means that experiments are being launched.