Similar to the process outlined in this doc for index builds

https://docs.mongodb.com/manual/tutorial/build-indexes-on-replica-sets/

It would be beneficial if the same process can be allowed in Atlas for running database maintenance processes such as dropping large databases that are no longer needed or to purge out old data using this method without impacting the primary.

Bulk Data can be purge out easier using the rolling maintenance method for example when implementing TTL index. If prior to implementing TTL index there is a lot of data to purge out that meets the TTL index criteria, this can be done prior to creating the index.