Issue #9425: file system encoding is not always utf-8

Fix a regression introduced in test_sys by r83778.

Lib/test/test_sys.py

@@ -509,10 +509,7 @@ class SysModuleTest(unittest.TestCase):
         p = subprocess.Popen([sys.executable, "-c", code], stderr=subprocess.PIPE)
         stdout, stderr = p.communicate()
         self.assertEqual(p.returncode, 1)
-        self.assertIn(
-            br"UnicodeEncodeError: 'utf-8' codec can't encode character "
-            br"'\udcff' in position 7: surrogates not allowed",
-            stderr)
+        self.assertIn(br"UnicodeEncodeError:", stderr)
 
     def test_sys_flags(self):
         self.assertTrue(sys.flags)