Orphaned adhoc tasks for refresh_mod_calendar_events_task in faildelay loop

This is in the cron log:

refresh_mod_calendar_events_task,Invalid course module ID

But this fails and then goes in circles forever. It should instead just emit an error and then not throw an exception and then get cleaned up.

I suspect this happens when a calendar is modified and then an activity maybe deleted quickly and the adhoc tasks get executed out of order.

Note: there is a whole class of different core bugs with adhoc tasks going in circles failing forever. See MDL-68768