typed_endpoint: Handle "dataclass_type" pydantic errors.

This commit adds error message for handling "dataclass_type"
pydantic errors. An example when this occurs is when group
setting value passed to update the setting is invalid like group
ID is passed directly and not in an object with "new" field and
other invalid values as well.

zerver/lib/typed_endpoint.py

@@ -316,6 +316,7 @@ def parse_view_func_signature(
 ERROR_TEMPLATES = {
     "bool_parsing": _("{var_name} is not a boolean"),
     "bool_type": _("{var_name} is not a boolean"),
+    "dataclass_type": _("{var_name} does not have the expected format"),
     "datetime_parsing": _("{var_name} is not a date"),
     "datetime_type": _("{var_name} is not a date"),
     "dict_type": _("{var_name} is not a dict"),

zerver/tests/test_user_groups.py

@@ -1319,6 +1319,11 @@ class UserGroupAPITestCase(UserGroupTestCase):
         leadership_group.deactivated = False
         leadership_group.save()
 
+        # Test updating with value not in the form of GroupSettingChangeRequest
+        params[setting_name] = orjson.dumps(support_group.id).decode()
+        result = self.client_patch(f"/json/user_groups/{leadership_group.id}", info=params)
+        self.assert_json_error(result, f"{setting_name} does not have the expected format")
+
     def test_update_user_group_permission_settings(self) -> None:
         hamlet = self.example_user("hamlet")
         check_add_user_group(hamlet.realm, "support", [hamlet], acting_user=hamlet)