Use diagnosed CO2 concentrations

[duncanwp]

The difference in CO2 concentrations that are used to derive the cumalitve emissions in the idealised CO2 scenarios will include ocean uptake, while the cumalitve emissions used for the historical and future scenario runs won't. This shouldn't make a huge difference but could be more consistent. It would probably be easiest to use the CO2 concentrations from the historical runs instead of cumulative emissions (although apparently input4MIPs includes concentrations too so I should check there first).