There are cases where a user would like the broker to automatically restart under specific conditions/events.

One case is to allow transparent handling of DBMS maintainance while using a JDBC journal: in such cases, most JDBC driver will just cause the broker JDBC connection to die, causing a shutdown of the broker.
It would be better, instead, to restart and await until the JDBC connectivity is restored (ideally) or just retry restarting allowing some think time to allow the DBMS to be available again.
A similar scenario could happen with NFS storage although NFS allow handling more transparently this while leaving the broker up and running.