The distillability problem revisited