Internal ops observability & alerting

[JSONbored]

Problem: No dashboards or alerts exist for the internal operations team to monitor active rented loops.

Area: Observability

Proposal: Wire dashboards and alerts for every active rented loop into the existing self-host observability stack.

Deliverables:

• Live internal dashboard panel(s) and alert rules for anomalous loop behavior.

Acceptance criteria:

• An internal operator can see every currently active rented loop and its status at a glance.

Boundaries:

• Blocked on Per-tenant execution sandboxing #4793.

---

Part of #4778.

[linear-code]

JSO-3480