added a comment - - edited
@[~jh194] I never counted, and now looking at the code I see (a) some optimization potential (e.g. redundant dequeue(event) calls and some requeue where enqueue suffices) and (b) that the operation I thought would trigger most requeues actually consists of "dequeue(event); time calculation; enqueue;" (but as I understand the HeapEventQueue, requeue is implemented just like dequeue(ev) and enqueue in sequence).
Anyway, there are 8802-9202 events in the queue (in steps of 100, resulting from different model properties and resulting in the different numbers of simulation steps; all runs were with same initial state and same rand seed (no replications), thus points in one "row" in the diagram probably belong to the same setting). Can we extract the number of requeues or dequeue(event)s from the difference of Calendar... and CalendarReQueue (or MList & MListRe now that the latter is working)? I guess writing an event queue wrapper counting the number of calls to each op is easier than guessing (but I can vouch that there is single dequeue() without event per simulation step