/jobs/:job/partitions

page usually gets stuck if the job has thousands of partitions. Maybe it should lazy-load partitions so we can start backfill on recently failed partitions. Not sure if there's any ticket for it. Happy to open one.