Single-shot k8s status for a tenant, backed by the k8s control plane — covers provisioning, readiness, and health, not just the initial provisioning phase. Lets the frontend tell "still provisioning" apart from a genuine failure instead of guessing from how long a direct pod ping has been failing (see wait_for_tenant_instance_provisioning for the equivalent polling loop used synchronously during creation itself).

Response

{"tenant_id": str, "status": "provisioning" | "healthy" | "suspended" | "failed" | "terminating" | "unknown"}

suspended means the tenant is scaled to zero with its data intact. It will not come up on its own — the client should offer a restore action (POST /{tenant_id}/restore) rather than keep polling.