Do not create two CO2 tracers

[gold2718]

• Only create one total CO2 tracer when a CO2 chemistry is running and co2_cycle is on
• When sharing total CO2 between co2_cycle and chemistry, do not create duplicate CO2 diagnostic fields
• The co2_cycle code will use the chemistry CO2 tracer (if it exists) as it's total CO2 quantity. The other tracers (for land, ocean and fossil fuel) are still locally created.

Addresses #1580

[gold2718]

While I tried to make this change relatively robust, I think there are a couple of weak points.

• The current solution suggestion is dependent on chem_register being called before co2_register.
• I am not currently making sure that CO2 is not on the non-transported list. I am not sure what to do about that potential case (i.e., throw an error).

[gold2718]

At a meeting about this, we decided that the diagnostic name for CO2 should be CO2 if the chemistry has CO2 as an active constituent and CO2esm if the constituent is created by the co2_cycle module.

However, I think it we would end up with two output fields for the same quantity. Currently, we already output all 3D constituent fields in diag_phys_writeout_dry (I think that is what is happening on line 972).

I think the same issue would affect the surface flux since those are also output using sflxnam in diag_srf.

Did I miss something or forget a detail? @klindsay28, @fvitt, @jimmielin, @nusbaume

[gold2718]

I can think of a couple of ways to implement the change above, however, we are not going to do this for NorESM (as among other things, it complicates post-processing and analysis) so I leave it up to someone from the CAM group to pick this up.

Some ideas for implementing CO2ems

• Change the name of the constituent from CO2 to CO2ems in co2_cycle.F90. I think this would effect the change but I'm not sure there is nowhere in CAM that would still look for or try to set CO2.
• Add a new array at the top of co2_cycle.F90. This could be called c_diag_names which would be the same as c_names except for CO2 ==> CO2ems. This new array would be used for addfld and add_default calls in co2_cycle and for outfld calls in cam_diagnostics.

[fvitt]

While I tried to make this change relatively robust, I think there are a couple of weak points.

• The current solution suggestion is dependent on chem_register being called before co2_register.
• I am not currently making sure that CO2 is not on the non-transported list. I am not sure what to do about that potential case (i.e., throw an error).

Probably should error out if CO2 is not transported. CO2 is a long-lived tracer and may never happen.

[gold2718]

Probably should error out if CO2 is not transported. CO2 is a long-lived tracer and may never happen.

Is this only if co2_cycle is on? It seems to me that it should also error out for GHG chemistry (which is a bit beyond the current scope of this PR).

[fvitt]

Probably should error out if CO2 is not transported. CO2 is a long-lived tracer and may never happen.

Is this only if co2_cycle is on? It seems to me that it should also error out for GHG chemistry (which is a bit beyond the current scope of this PR).

Yes only if co2_cycle is active.