Hi RunningRoach,
I would consider the replay log to be another queue for the purposes of consuming resources against the limits you identify.
Our documentation often refers to non-durable queues as temporary queues, which are a subset of temporary endpoints. If you search based on “temporary queues”, you will find that only more recent releases support replay with non-durable/temporary queues. Note that a temporary endpoint is bound to the consumer that established them and is removed when that consumer disconnects. You cannot bind an alternate consumer to a temporary endpoint like a temporary queue.