Can confirm this works, I have a review session with Databricks next week some time to green light the process we have used, but seems to be working well enough for now. Only amendment to that design proposal was storing flows on blob storage as opposed to the docket image.