Agent._regenerate_code_after_error expects InvalidLLMOutputType exception, but no code ever raises it

[sevdokimov]

I noticed that Agent._regenerate_code_after_error uses a different prompt when handling the InvalidLLMOutputType exception compared to other exceptions. However I don't see where InvalidLLMOutputType is ever raised in the code. When the code returns a wrong result, ResponseParser raises InvalidOutputValueMismatch or NoResultFoundError.
Should InvalidOutputValueMismatch and NoResultFoundError inherit from InvalidLLMOutputType?