Convert ovens to a persistent work queue, budgeted to 0.25ms by default In the 1000 oven test case this caps out at processing around 40-50 ovens a frame