LGTM but please add a test to show what happens when the message containing the schedule is deleted, aged out, purged etc.

Extended the test (TestJetStreamClusterScheduledIntervalMessage) to confirm the scheduling stops when the scheduling subject is removed. Other pre-existing tests already check for other cases, like "overwriting" a schedule by publishing a different schedule for a given subject.

Purging and aging out isn't explicitly tested I don't think, but probably doesn't need to be as the schedule itself needs to exist (be loaded from the store) in order for a message to be scheduled, if it doesn't exist (removed by any means) it's deactivated and removed from the hash wheel.