fix(quotas): unifier le décompte IA (BYOK, rollback) et combler les fuites

Centralise la réserve via ai-quota, corrige admin unavailable (-1), brancher les routes sans quota et le host-pays brainstorm, avec usage-meter élargi, noms de clusters, MCP et ajustements dashboard/insights.

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

memento-note/lib/ai/services/memory-echo.service.ts

@@ -16,6 +16,7 @@ import {
   buildMemoryEchoInsightPrompt,
   getMemoryEchoInsightFallback,
 } from '@/lib/ai/memory-echo-i18n'
+import { withAiQuota } from '@/lib/ai-quota'
 import {
   SEMANTIC_SIMILARITY_FLOOR_CLIP,
   SEMANTIC_SIMILARITY_FLOOR_DEMO,
@@ -463,14 +464,20 @@ export class MemoryEchoService {
       return null // All connections already shown
     }
 
-    // Generate AI insight in user's language
+    // Generate AI insight in user's language (quota: reformulate, with rollback on failure)
     const userLanguage = await detectUserLanguage()
-    const insightText = await this.generateInsight(
-      newConnection.note1.title,
-      newConnection.note1.content,
-      newConnection.note2.title,
-      newConnection.note2.content || '',
-      userLanguage,
+    const insightText = await withAiQuota(
+      userId,
+      'reformulate',
+      () =>
+        this.generateInsight(
+          newConnection.note1.title,
+          newConnection.note1.content,
+          newConnection.note2.title,
+          newConnection.note2.content || '',
+          userLanguage,
+        ),
+      { lane: 'chat' },
     )
 
     // Store insight in database