feat: connection rate and chain latency metrics

Add RateTracker (rolling deque, events/sec) and LatencyTracker (circular
buffer, p50/p95/p99 in ms) to the Metrics class.  Both are recorded in
_handle_client and exposed via summary(), to_dict(), /status, and /metrics.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

docs/CHEATSHEET.md

@@ -164,7 +164,22 @@ python -m pstats ~/.cache/s5p/s5p.prof      # container profile output
 ## Metrics Log
 
 ```
-metrics: conn=142 ok=98 fail=44 retries=88 active=3 in=1.2M out=4.5M up=0h05m12s pool=42/65
+metrics: conn=1842 ok=1790 fail=52 retries=67 active=3 in=50.0M out=1.0G rate=4.72/s p50=198.3ms p95=890.1ms up=1h01m01s pool=42/65
+```
+
+## Metrics JSON (`/metrics`)
+
+```bash
+curl -s http://127.0.0.1:1081/metrics | jq .
+```
+
+```json
+{
+  "connections": 1842,
+  "success": 1790,
+  "rate": 4.72,
+  "latency": {"count": 1000, "min": 45.2, "max": 2841.7, "avg": 312.4, "p50": 198.3, "p95": 890.1, "p99": 1523.6}
+}
 ```
 
 ## Troubleshooting