We expect the queue to remain at or near a depth of zero at all times. Batched merges ensure this is possible by increasing the maximum merge capacity.

When we did not have batched merges it would often take a week or more to merge code. This resulted in our adding a lot of complicated logic to the submit queue to order things correctly.

We want to favor maximizing the chance that a batched run can merge all the things, so eliminate reordering:

• based on priority labels: https://github.com/kubernetes/test-infra/blob/master/mungegithub/mungers/submit-queue.go#L1114
• based on retest not required labels: https://github.com/kubernetes/test-infra/blob/master/mungegithub/mungers/submit-queue.go#L1286

(we should still skip retesting with the retest not required labels, but we should only add things to the end of the queue)