Alerting when new provider assignment eats up all the iops of a shard

[willscott]

We recently had Pinata join as an elastic IPFS provider and it got assigned to an already busy shard. this resulted in the shard not being able to keep up with it's allocated read workload.

• make sure alerting is able to notify us that the iops aren't keeping up with needs on a shard
• triage if there's something smarter we can do with assignment of new providers to shards