CESM errors for 'andreas' algo

[jbusecke]

I just narrowed down an issue I did not address earlier. When computing the fluxes with different algos for the paper appendix, it seems that the andreas algo always leads to an error resulting in a killed workers. I am not sure what exactly is happening but it smells like some issue on the aerobulk side, likely due to some out of range value, or a lack of convergence?

• For the paper I will just drop this algo from the CESM analysis, our results are conclusive enough without it.
• In general this just again points us to work on better error handling for aerobulk-python Better error handling xgcm/aerobulk-python#4

[jbusecke]

Quick addition, this seems to also apply to both COARE algorithms.