The ventilatory response to modified rebreathing is unchanged by hyperoxic severity: implications for the hyperoxic hyperventilation paradox.