The jstest added by SERVER-102999 checks a procedure that could be used by the backup team to restore a cluster faster than right now by creating cursors with a find command from a router and then consuming them on each shard simultaneously. By doing it this way each cursor will take into consideration ownership. However, in Atlas, authentication is enabled, and more importantly, authentication between processes which could make the procedure fail with the error:

failed to run getMore command: (Unauthorized) cursor id 2101070483501344165 was not created by the authenticated user 

This happens because the user used to create the cursor by the router is __system, so, if you use a different user, even if is the same user that issued the command on the router, the getMore command to exhaust the cursor would fail.