I'm glad to hear that updating your nginx configuration helped with the timeouts! As long as you are triggering the workflow in an asynchronous manner, it shouldn't have any issues running for multiple hours. I think the ceiling is something like 24 days.
That said, it's still possible to be bottlenecked by data transfers and memory allocation. It would probably be most helpful to take a look at your container logs when the workflow fails. If you do want to hop on a call, we host Office Hours every Tuesday and Thursday!