SchedulerOptions.tickMs, plus
twice leaderTtlMs with a leader lease); a slot interrupted by a crashed runner is
retried as itself instead:
'skip'(default): run none of them and wait for the next slot after now.'run-once': run one occurrence (missed: true,scheduledAt= the first missed slot), then wait for the next slot after now.'catch-up': run every missed slot in order, each with its ownscheduledAt, up to DurableTaskOptions.maxCatchUp; later missed slots are skipped (logged).
delay task counts as one slot: 'skip' drops a reminder that was due
while the bot was down, the other two run it once.