I wonder how well their results reflect actual battery failure scenarios, where the battery fails not due to heating supplied from the outside (as in the displayed experiment), but due to internal heating from a short circuit / isolation failure. Could make a difference.