Agents cannot start more than 10 job instances at once
When starting more than 10 job instances on the same agent, instances started after the 10th one will immediately come to a stop and further stop job instances won't work either.
This seems to be related to APScheduler defaults of using a thread pool of 10 workers to fire scheduled tasks. Increasing the amount of workers or switching to a new way of managing started jobs (such as using the
ProcessPoolScheduler) is required to handle a larger amount of tasks per agent at once.