Add tests for errors

Currently, we mostly test the successful execution of commands.  We could add more tests for the error cases, e. g. invalid data, API calls for missing keys or users, wrong roles or no authentication at all, wrong system state, etc.