This is the leftover from the days we needed advanced heuristics that had to know more about allocation times in the regions. Currently, only Traversal needs one of those sequence numbers, yet we always carry four of them.

Seqnum machinery increments a global variable on the locked allocation path. Doing excess things there inflates the time thread spends in its critical section, especially if it runs into conflict from another thread allocation. We should consider dropping the three unused seqnums, and conditionalize the leftover one for Traversal.