Get Containers Health

Health of this workspace's MCP workloads.

The manager also has a route that answers for every workload it runs, which is the wrong thing to hand a caller: its rows carry service names and container ids belonging to other workspaces. The instance list comes from the workspace-scoped service instead, and only those ids are asked about, so the answer cannot name a workload the caller is not entitled to see.

A workload the manager reports as unhealthy, has never heard of, or cannot be reached about is a fact about that instance, not a failure of the request: each is reported per instance and the endpoint still answers 200.