Metrics: don't count the dashboard's own /api/v2/metrics polling (#135)

The ops dashboard polls the metrics endpoint every few seconds; counting those hits
just shows the monitor watching itself and inflates the inbound totals. Skip the
'metrics' category in record_inbound (so it's excluded from inbound_total too), and
hide it in the dashboard's inbound list defensively.

metrics.py

@@ -97,6 +97,10 @@ def classify_inbound(path: str, base: str = "") -> str:
 
 def record_inbound(category: str, status) -> None:
     """Count one inbound request in ``category``, bucketed by status class."""
+    # The metrics endpoint is polled by the dashboard itself — counting it would just
+    # show the monitor watching itself, so it's excluded from traffic entirely.
+    if category == "metrics":
+        return
     try:
         bucket = _status_bucket(status)
         with _lock:

tests/test_metrics.py

@@ -60,6 +60,17 @@ def test_record_inbound_buckets_by_status():
     assert snap["inbound_total"] == 4
 
 
+def test_metrics_endpoint_polling_is_not_counted():
+    # The dashboard polls the metrics endpoint; that self-traffic must be excluded.
+    m = _fresh()
+    m.record_inbound("metrics", 200)
+    m.record_inbound("metrics", 404)
+    m.record_inbound("api:place", 200)
+    snap = m.snapshot()
+    assert "metrics" not in snap["inbound"]
+    assert snap["inbound_total"] == 1
+
+
 def test_record_outbound_by_source_and_outcome():
     m = _fresh()
     m.record_outbound("history_fetch", "ok")