When splitting tests into subtasks, if a test doesn't have any test history we don't know enough information to override the timeouts. So, we shouldn't do that.