MoV update to run smoothly with diverse datasets

[lee1043]

This PR is to make the code more flexible to run it for other datasets than CMIP. In particular for COBE2. @kristinchang3 reported an issue with CODE2, which was resulted from non-standard variable name in COBE2 when it was treated like a model data. (sst in COBE2 rather than ts). The variable name ts was used as a part of decision to apply landmask. The change in this PR allows 'sst' in addition. Further generalization might be needed in the future.

[kristinchang3]

Thanks so much @lee1043 for figuring out this fix! I just ran the new driver script successfully. I will paste the result below. I noticed that there are some extra plots that end up being produced that don't make sense. For example, there are two teleconnection plots for the obs dataset, one for sst and one for ts. The same thing occurs for the monthly plot as well. I will paste these below for your review. Is it easy to exclude those plots from being produced?

[lee1043]

Hi @kristinchang3, thanks for following up. That is interesting that the individual plot for HadISST is not matching to the summary plot. Any chance they were generated from earlier production before the change is made? The new files from obs should include 'ts' instead of 'sst' (which I corrected in this PR), so the file name is consistent to the variable name in the dataset. Could you rename the existing folder to something else, rerun the process, and check the image files you get? Sorry if you have don't this already (if so just let me know), but I just wanted to make sure before jumping in to the code.

[kristinchang3]

Oh good catch @lee1043 you were right - I had some images saved in the same case directory from a previous run. I just reran and everything looks good! Thanks again!