On the impact of recent developments of an atmospheric general circulation model on the simulation of CO2 transport