feat(wechat): add admin-pluggable LLM intent router separate from H5.

Give WeChat MP its own chat.general→page.generate LLM refinement layer with memindadm toggles, shadow mode, and canary openids so service account routing stays independent of the H5 chatIntentRouter.

Co-authored-by: Cursor <cursoragent@cursor.com>

server.mjs

@@ -280,6 +280,7 @@ let skillRuntimeConfigService = null;
 let systemDisclosurePolicyService = null;
 let agentCodeRunPolicyService = null;
 let wechatScheduleLlmConfigService = null;
+let wechatIntentRouter = null;
 let mindSpace = null;
 let mindSpaceAssets = null;
 let mindSpaceAudit = null;
@@ -471,6 +472,8 @@ async function bootstrapUserAuth() {
       memorySessionServices.agentCodeRunPolicyService;
     wechatScheduleLlmConfigService =
       memorySessionServices.wechatScheduleLlmConfigService;
+    wechatIntentRouter =
+      memorySessionServices.wechatIntentRouter;
     conversationMemoryService =
       memorySessionServices.conversationMemoryService;
     memoryV2 = memorySessionServices.memoryV2;
@@ -530,6 +533,7 @@ async function bootstrapUserAuth() {
         scheduledTaskService,
         sessionSnapshotService,
         wechatScheduleLlmConfigService,
+        wechatIntentRouter,
         llmProviderService,
         chatIntentRouter,
         systemDisclosurePolicyService,