Serrin-type overdetermined problems: An alternative proof