fix: test failure due to incorrect api usage

- test provides internal dict structure directly but provides a
  non-string input for "default" whereas normally these values are
  provided from XLS/X or MD as strings.
- the failure is at the function `has_last_saved` which looks for the
  length of a value but that is a TypeError for `int` - probably there
  are other places that would expect strings as well.

tests/test_unicode_rtl.py

@@ -39,7 +39,7 @@ def test_smart_quotes(self):
                         "type": "integer",
                         "name": "my_default_is_123",
                         "label": "my default is 123",
-                        "default": 123,
+                        "default": "123",
                     },
                 ],
                 "choices": [