Highlight dummy shards in collection cluster web UI #346
Description
If user creates a snapshot and it fails, that leaves behind a dummy shard which silently rejects read requests and is only healed when a upsert request is made. But this upsert might never happen and the replica can remain forever in a "dummy shard" stage which affects availability and throughput.
We should have a way to at least show to the user that one of their replicas are dummy and it's a problem. So we may extend our collection cluster visualization to show this. It might require changes in the core API or using the new cluster telemetry endpoint.
