feat(cluster): Rolling Update — Secondary-first upgrade orchestration (v1.2.3)

POST /cluster/rolling-update startet den gestaffelten Upgrade-Prozess:
1. Secondary via mTLS /agent/cluster/trigger-update anstoßen
2. /agent/cluster/version pollen bis Secondary Version-Flip zeigt (max 10 min)
3. Primary self-upgrade via systemd-run (identisch zu /system/upgrade)

State wird in /var/lib/edgeguard/rolling-update-state.json persistiert:
Phasen: updating-secondary → waiting-secondary → updating-primary.
"done" wird nicht geschrieben — Prozess stirbt beim Upgrade. UI erkennt
Abschluss via /system/health version-flip (analog Single-Node-Upgrade).

UI: UpdateBanner erkennt Cluster-Modus (/cluster/status mode="cluster")
und tauscht den "Install now"-Button gegen "Rolling Update (Cluster)" aus.
Multi-Step-Modal zeigt die drei Phasen; ab updating-primary wechselt der
Client auf /system/health polling.

Aggregator.PostPeer: neuer einzel-POST-Helper für mTLS-trigger-update.
WithVersion(): ClusterHandler bekommt Binary-Version für /agent/cluster/version.

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

management-ui/src/i18n/locales/en/common.json

@@ -629,6 +629,7 @@
       "node": "Node",
       "status": "Status",
       "role": "Role",
+      "pgRole": "PG role",
       "apiUrl": "API URL",
       "configHash": "Config hash",
       "version": "Version",
@@ -641,6 +642,30 @@
       "uptime": "Uptime",
       "fetchMs": "Fetch"
     },
+    "pgRole": {
+      "standalone": "standalone",
+      "primary": "primary",
+      "standby": "standby"
+    },
+    "vipCard": {
+      "title": "High Availability (VIP / Keepalived)",
+      "vipAddress": "VIP address",
+      "vipAddressHelp": "Virtual IP address that moves between nodes (e.g. 89.163.205.10)",
+      "vipInterface": "Network interface",
+      "vipInterfaceHelp": "Interface to bind the VIP on (e.g. eth0)",
+      "vipAuthPass": "VRRP auth password",
+      "vipAuthPassHelp": "Max. 8 characters — Keepalived limit. Same on all nodes.",
+      "vrrpRouterId": "VRRP router ID",
+      "vrrpRouterIdHelp": "Must be unique in the subnet (1–255). Default: 51.",
+      "saveBtn": "Save & reconfigure Keepalived",
+      "saved": "VIP settings saved",
+      "saveFailed": "Failed to save",
+      "hintTitle": "Next steps after saving",
+      "hintPrimary": "On primary: edgeguard-ctl cluster-init-replication",
+      "hintStandby": "On secondary: edgeguard-ctl cluster-setup-standby <primary-ip>",
+      "hintKeepalived": "Keepalived on both nodes: sudo systemctl enable --now keepalived",
+      "hintFailover": "On failover: edgeguard-ctl promote (on the secondary node)"
+    },
     "loadTitle": "Per-node resources (mTLS aggregator)",
     "loadEmpty": "No node resources available — agent listener unreachable?",
     "certCardTitle": "Cluster TLS certificates",
@@ -841,7 +866,14 @@
     "stepDownload": "Download packages",
     "stepInstall": "Install",
     "stepRestart": "Service restart",
-    "stepVerify": "Verification"
+    "stepVerify": "Verification",
+    "rollingUpdate": "Rolling Update (Cluster)",
+    "rollingConfirmTitle": "Start Rolling Update?",
+    "rollingConfirmDesc": "The secondary node ({{secondary}}) is updated first, then this primary. No downtime for proxied traffic during the secondary phase.",
+    "rollingRunning": "Rolling update in progress…",
+    "rollingStepSecondary": "Updating secondary ({{fqdn}})",
+    "rollingStepWaiting": "Waiting for secondary restart",
+    "rollingStepPrimary": "Updating primary (this node)"
   },
   "wg": {
     "title": "WireGuard",