This is slightly difficult to answer without knowing your architeture and setup. I am assuming you have a dedicated Premium Capacity and my response is geared towards this assumption.
Refreshes along with Subscriptions is considered backend processes while users viewing content and interacting with reports is considered frontend processes.
As long as resources are not constrained, review your premium capacity metrics app, I would believe what you outlined would work (assuming the refresh duration doesnt take an hour). However, if your capacity is 'peaking', Microsoft prioritizes frontend over backend processes and queues the backend.
To add more complexity, you are only allowed so many concurrent refreshes per capacity p node, a P1 will allow 6 concurrent refreshes at a time.
If I was tasked to figure this out, I would study the refresh history of the dataset to establish a baseline on how long it normally takes to complete. I would then identify what other datasets are scheduled at the same time to see if there will be any concurrency issues. As a precaution, I would also review the metrics app to see when my organizations peak usage is. I would also review I would then schedule the subscription to myself and monitor it for a week, seeeing when I receive the email versus knowing when I scheduled the email.
Hopefully this provides you the tools to help out.
