fix: return clean errors and refund quota on AI provider failures

AI routes had no consistent error handling — a provider failure (e.g.
OpenRouter returning a degraded-model error) crashed the route as an
uncaught 500 and, worse, still charged the user's monthly aiCall
quota for a request that never succeeded.

Adds withAiQuota()/aiErrorResponse() (lib/ai/ai-error.ts) and applies
them across all 12 AI generation routes: charges the quota, runs the
AI call, refunds the credit and returns a clean user-facing message
if it throws. Frontend needs no changes — existing dialogs already
toast whatever `error` string comes back.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

apps/web/lib/tiers.ts

@@ -69,6 +69,20 @@ export async function checkAndIncrementTierLimit(
   }
 }
 
+/**
+ * Refunds one aiCall credit for the current month. Call this when an AI
+ * request failed after the quota was already charged (e.g. provider error)
+ * so users aren't billed against their limit for a call that never succeeded.
+ */
+export async function refundAiCall(userId: string): Promise<void> {
+  const month = currentMonth();
+  await db.execute(sql`
+    UPDATE user_usage
+    SET ai_calls_used = GREATEST(ai_calls_used - 1, 0)
+    WHERE user_id = ${userId} AND month = ${month}
+  `);
+}
+
 export async function incrementUsage(
   userId: string,
   key: LimitKey,