This is a ~feature request related to recent work done by @jewettaijfc in #2648, #2647, and #2600.
When running a large batch, particularly when the simulations are fast, I find that the time to download results is a substantial fraction of the total time time (from submission -> results on local machine).
Is it possible to have results files begin to download as soon as an individual simulation completes? From my end it looks like each simulation is handled independently server-side, so I can interact with a completed simulation through the web GUI while others in the batch run... but it may be tricky to having a single python kernel manage server interactions to monitor batch progress and simultaneously download?