Observability: Prometheus metrics, log panel, runbook

- GET /api/metrics renders Prometheus text (equity, P&L, per-strategy
  equity/pnl/allocation/halted, info) with no client library.
- In-memory ring log handler (app/logs.py) exposed at GET /api/logs and a
  dashboard "Server logs" panel; Runtime mirrors bot events into the logger.
- docs/runbook.md covers deploys, the kill switch, resetting paper state and
  going live. Adds tests for metrics rendering and the log ring.

AGENTS.md

@@ -136,6 +136,17 @@ the single and multi start paths are covered) and surfaced via
 call `check_live_ready` first. `LIVE_CONFIRMATION` must equal
 `"I UNDERSTAND THE RISK"` — keep it a phrase, not a boolean.
 
+## Observability
+
+- `app/metrics.py` renders Prometheus text at `GET /api/metrics` straight from
+  the manager status — no client library. Keep new series out of hot loops.
+- `app/logs.py` keeps a small in-memory ring of log records, exposed at
+  `GET /api/logs` and shown in the dashboard. `Runtime._log_event` mirrors bot
+  events into the logger so the panel is useful (uvicorn's own loggers do not
+  propagate to the root handler).
+- `docs/runbook.md` is the operations guide; update it when deploy or
+  recovery steps change.
+
 ## Risk controls & notifications
 
 - `app/engine/risk.py` (`RiskManager`) tracks daily loss, drawdown, position cap