Enhance error handling in AIEngine and update ResourceDebugOverlay

- Added error type handling in AIEngine for better error categorization during model configuration and execution.
- Updated _handle_error method to accept and log error types.
- Improved ResourceDebugOverlay to silently ignore 404 responses from the metrics endpoint, preventing unnecessary logging and retries.
- Refactored authStore to utilize fetchAPI for automatic token handling and improved error logging without throwing exceptions.

frontend/src/components/debug/ResourceDebugOverlay.tsx

@@ -128,7 +128,7 @@ export default function ResourceDebugOverlay({ enabled }: ResourceDebugOverlayPr
         headers['Authorization'] = `Bearer ${token}`;
       }
       
-      // Silently handle 404s and other errors - metrics might not exist for all requests
+        // Silently handle 404s and other errors - metrics might not exist for all requests
       try {
         const response = await nativeFetch.call(window, `${API_BASE_URL}/v1/system/request-metrics/${requestId}/`, {
           method: 'GET',
@@ -136,6 +136,11 @@ export default function ResourceDebugOverlay({ enabled }: ResourceDebugOverlayPr
           credentials: 'include', // Include session cookies for authentication
         });
         
+        // Silently ignore 404s - metrics endpoint might not exist for all requests
+        if (response.status === 404) {
+          return; // Don't log or retry 404s
+        }
+        
         if (response.ok) {
           const responseData = await response.json();
           // Extract data from unified API response format: {success: true, data: {...}}