A recent change to reset unaffiliated bitmaps during the concurrent cleanup phase has caused a performance regression. Likely because concurrent cleanup only uses one thread (at this time). This issue is to revert the regression causing change while we work on a preferred solution.