This reduces the number of active queues, which reduces the request rate to be in line with the reduced number of instances.
See https://github.com/m-lab/etl/pull/959
Coverage remained the same at 60.479% when pulling f5356ab0dbfae9207515f80dfd6c45daa939fb09 on sandbox-half-queues into 03d9f4faf6bb588f1220f69d0395c9d4ae01e2a1 on master.
This reduces the number of active queues, which reduces the request rate to be in line with the reduced number of instances. See https://github.com/m-lab/etl/pull/959
This change is