Persist the next trigger before starting scheduled work

When scheduled work runs in an interruptible runtime and uses a one-shot timer, persist the next trigger before beginning long work. Scheduling only in cleanup can stop future runs because runtime termination skips cleanup. Recovery should recreate a missing trigger while preserving an existing overdue trigger; blindly replacing it can postpone work that has not yet been dispatched. Track interrupted runs separately from durable queued records so a partial run does not erase already saved data. This prevents avoidable scheduling gaps, but does not guarantee execution while the host application is closed.

Applies when: A short-lived background runtime performs scheduled collection or delivery. Keywords: scheduled work one shot timer interruption retry overdue recovery persistence