Please invent VOID-Function Calling with Assistants (action_required flag `None`)

This is just feedback to the OpenAI team.
While I do understand the concept and common use of function calling (fc) in association with Assistants, but there is still another powerful application of fc without assistant (basic fc), that cannot be applied when using the same in the context of assistants. In basic fc we can use the function to fix the model to obey a specific desired output format (e.g. JSON) in response (packaged in the function arguments) to a prompt. However, because, in fc with assistants, the assistant always turns into a state of action required and waiting, while we can encapsulate the quasi interim response again packaged in the arguments, we cannot complete. We practically could manage that the assistant packages its response in a certain format via arguments and get even the result of its action encapsulated there, but we can not simply continue on the same thread because the run is waiting for action and does not complete. Indeed, there seems to be an ugly solution to this, by just acting like submitting a dummy response to the assistant as output hoping that the assistant would complete but even this can get extremely challenging because of the asynchronous tact of the assistant system if our program would like to send the next prompt in the thread. Practically, it would be wonderful if you could implement a flag for fc, by which one could use it with assistants, in a mode where the flag indicates that this fc cannot require action. A kind of no-action function flag. In this way, we could use fc beyond its present common application for calling functions/external operation/response ... / complete to assistant in those many situations where we explicitly need a simpler scheme, namely only function/external operation/[no-response]/complete. This can be called void-function` calling. This is really a hurdle for developing a spectrum of intelligent products.