.partial() sets arguments that are the same for every copy; .expand() sets the ones that vary:
count_status.partial(status="failed_attempt").expand(depot=depots())
Give expand two arguments and Airflow makes every combination: expand(depot=["leeds", "york"], status=["delivered", "failed_attempt"]) makes four copies. To pair values instead, pass a list of dicts to expand_kwargs, one dict per copy:
is_late.expand_kwargs([{"depot": "leeds", "cutoff": "06:00"}, {"depot": "york", "cutoff": "06:30"}])
The list can come from a task, which is the usual case. Classic operators map the same way, with the operator arguments in partial: BashOperator.partial(task_id="archive").expand(bash_command=commands()). task_id, retries and pool can never be mapped. For TaskFlow tasks, put operator arguments such as retries in the decorator.
One copy per depot, each with its own cutoff time.