On the practical identifiability of a two-parameter model of pulmonary gas exchange