We are using Herd Avoidance to prevent users from executing simultaneous executions of the same task. However, if there's an exception in the task and we retry from within the task, the retried task ends up hitting the herd avoidance as well. Is this working as designed? Shouldn't retried tasks be able to avoid the herd avoidance?