I have an MLflow model registered in Azure and I deploy it through the serving MLflow UI as an endpoint that is stood up until I turn it off. The endpoint is only needed to be c