On the global attractivity controversy for a delay model of hematopoiesis