Design Distributed Cron

Design a distributed cron service: users define recurring jobs (cron expressions with a time zone) and the system triggers each occurrence on a fleet of runners. Focus on firing every occurrence exactly once at the decision level even when schedulers crash or lose leadership, on missed-run and overlap policies, and on time zone and daylight saving behaviour. This is about time-based recurrence; a general task queue with delays and dependency graphs are separate questions.