perf: stats poll via lichtgewicht /api/stats i.p.v. /containers-dashboard

De frontend haalde CPU/mem stats op via het zware /containers-dashboard
endpoint (Podman call + os.walk + systemctl subprocesses per container).
Nu gaat de stats poll via een nieuw /api/stats endpoint dat alleen de
bestaande in-memory cache teruggeeft (<5ms vs ~400ms).

- app_containers.py: /api/stats endpoint toegevoegd (cache direct return)
- app_containers.py: _STATS_SHOWN_NAMES bijgehouden per dashboard call
  (filtert infra/management containers eruit op basis van _dashboard_source)
- containers.js: pollContainersDashboardStatsOnce() gebruikt /api/stats

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

control/app_containers.py

@@ -30,6 +30,7 @@ _PODMAN_API_BASE = None
 _STATS_CACHE_BY_NAME = {}  # name -> {"cpu": float|None, "mem_usage": float|None, "mem_perc": float|None}
 _STATS_CACHE_TS = None
 _STATS_POLLER_TASK = None
+_STATS_SHOWN_NAMES: set = set()  # namen van systemd-managed containers uit laatste dashboard call
 
 # --- EXEC SESSION CACHE (in-memory) ---
 _EXEC_SESSIONS = {}  # session_id -> _ExecSessionState
@@ -478,8 +479,23 @@ def init_containers_router(
             row["Status"] = (out or "").strip()
             dashboard.append(row)
 
+        # Bijwerken welke namen systemd-managed zijn (voor /stats filter)
+        global _STATS_SHOWN_NAMES
+        _STATS_SHOWN_NAMES = {
+            _norm_container_name((c.get("Names") or ["?"])[0])
+            for c in dashboard
+            if c.get("_dashboard_source") == "systemd"
+        } - {"?", ""}
+
         return dashboard
 
+    @router.get("/stats")
+    def stats_snapshot():
+        cache = _STATS_CACHE_BY_NAME
+        if _STATS_SHOWN_NAMES:
+            return {k: v for k, v in cache.items() if k in _STATS_SHOWN_NAMES}
+        return cache
+
     @router.get("/containers")
     def list_containers():
         # Ook hier ?all=true voor gestopte containers