chore: improve monitoring for fix: handle Firefox dynamic import error message i

**Monitoring fix:** Reordered error handling in `AppErrorBoundary.handleError` to check for stale chunk errors BEFORE sending to Sentry.

Previously, every stale chunk error was reported to Sentry and THEN auto-recovered via `window.location.reload()`. This created Sentry noise for errors that are transient and auto-recovered. Now the stale chunk check happens first — if detected, the page reloads immediately without sending to Sentry, since the user never sees the error.

This eliminates a class of Sentry noise: post-deployment stale chunk errors that are fully auto-recovered.

packages/twenty-front/src/modules/error-handler/components/AppErrorBoundary.tsx

@@ -22,6 +22,15 @@ export const AppErrorBoundary = ({
   resetOnLocationChange = true,
 }: AppErrorBoundaryProps) => {
   const handleError = async (error: Error | CustomError, info: ErrorInfo) => {
+    const isViteStaleChunkLazyLoadingError =
+      checkIfItsAViteStaleChunkLazyLoadingError(error);
+
+    if (isViteStaleChunkLazyLoadingError) {
+      window.location.reload();
+
+      return;
+    }
+
     try {
       const { captureException } = await import('@sentry/react');
       captureException(error, (scope) => {
@@ -36,13 +45,6 @@ export const AppErrorBoundary = ({
       // oxlint-disable-next-line no-console
       console.error('Failed to capture exception with Sentry:', sentryError);
     }
-
-    const isViteStaleChunkLazyLoadingError =
-      checkIfItsAViteStaleChunkLazyLoadingError(error);
-
-    if (isViteStaleChunkLazyLoadingError) {
-      window.location.reload();
-    }
   };
 
   const handleReset = () => {