feat(dataflow): live Data Flow graph API + PII catalog

api/dataflow.py: node-link landscape (generators->sources->CDC/Kafka->sinks,
HDFS->Iceberg, sources->curated_masked) with live overlays (movement run state,
CDC volume, Trino counts) and PII overlay. api/pii_catalog.py: column PII
classification via Trino information_schema (OM-ready). Endpoints /api/dataflow,
/api/dataflow/{id}/run, /api/pii.

api/cdc_consumer.py

@@ -180,6 +180,23 @@ async def cdc_consumer_loop() -> None:
                     pass
 
 
+def snapshot(minutes: int = 15) -> dict[str, Any]:
+    """Lightweight CDC snapshot for other modules (Data Flow graph)."""
+    cutoff = datetime.now(timezone.utc).timestamp() - minutes * 60
+    by_source: dict[str, int] = {}
+    total = 0
+    for c in _ring:
+        try:
+            if datetime.fromisoformat(c["ts"]).timestamp() < cutoff:
+                continue
+        except Exception:
+            continue
+        total += 1
+        by_source[c["source"]] = by_source.get(c["source"], 0) + 1
+    return {"connected": _state["connected"], "consumed": _state["consumed"],
+            "buffered": len(_ring), "window_total": total, "by_source": by_source}
+
+
 # ── Endpoints ────────────────────────────────────────────────────────────────
 @router.get("")
 async def list_changes(