When we throw staleConfig, we should generally have access to the sharding state since the shardId in the error represents the shard throwing the error. We should update these places to generate a stale config exception containing a shard ref, with a UUID when present. This will also require updating handlers of stale config exceptions to be okay with this change.