feat: add router health metrics to node_cache and node-guardian

- Add migration 042_node_cache_router_metrics.sql
- Node guardian now collects router health and sends in heartbeat
- City-service uses cached router_healthy from node_cache
- This allows NODE2 router status to be displayed correctly

services/city-service/repo_city.py

@@ -3373,7 +3373,7 @@ async def get_node_endpoints(node_id: str) -> Dict[str, str]:
 
 async def get_node_metrics(node_id: str) -> Optional[Dict[str, Any]]:
     """
-    Отримати розширені метрики ноди (включаючи Swapper).
+    Отримати розширені метрики ноди (включаючи Swapper та Router).
     """
     pool = await get_pool()
     
@@ -3385,7 +3385,9 @@ async def get_node_metrics(node_id: str) -> Optional[Dict[str, Any]]:
             swapper_models_total,
             swapper_state,
             router_url,
-            swapper_url
+            swapper_url,
+            router_healthy,
+            router_version
         FROM node_cache
         WHERE node_id = $1
     """, node_id)

services/city-service/routes_city.py

@@ -4086,13 +4086,29 @@ async def get_node_swapper_detail(node_id: str):
 async def get_dagi_router_health(node_id: str):
     """
     Get DAGI Router health status for a node.
-    Always returns 200 with status="down" if router is unavailable.
-    Uses node-specific router_url from node_cache.
+    First checks node_cache for cached router_healthy status (from node-guardian).
+    Falls back to direct health check if node is local (NODE1).
     """
     import httpx
     import time
     
-    # Get router URL from database (node-specific)
+    # First, try to get cached router health from node_cache
+    # This is populated by node-guardian which has direct access to the router
+    try:
+        metrics = await repo_city.get_node_metrics(node_id)
+        if metrics and metrics.get("router_healthy") is not None:
+            return {
+                "node_id": node_id,
+                "status": "up" if metrics.get("router_healthy") else "down",
+                "version": metrics.get("router_version"),
+                "agent_count": 0,  # TODO: get from node_cache
+                "latency_ms": None,
+                "source": "node_cache"
+            }
+    except Exception as e:
+        logger.debug(f"Failed to get cached router health for {node_id}: {e}")
+    
+    # Fallback: try direct health check (only works for NODE1 which is local to city-service)
     endpoints = await repo_city.get_node_endpoints(node_id)
     base_url = endpoints.get("router_url")