Testing for CUPiD workflow in CTSM

In order to make sure the CUPiD workflow works, we'd like to add some testing that you can both run CUPiD in a test that it completes without errors, and that it creates: datasets, notebooks, and plots. Since it's a machine PASS/FAIL it will NOT check that the plots are correct or accurate as that requires a human to look at them.

CTSM SystemTests:
- [ ] CUPiD runs and runs without errors
- [ ] Datasets are created (and compared to baselines and make sure they agree)
- [ ] Expected notebooks are created
- [ ] Plots are created as expected (this will check with CUPID_GEN_HTML==TRUE)

CTSM Tests using above to be put into either aux_clm or ctsm-sci
- [ ] Run ROF tests as well
- [ ] Run LND option
- [ ] Run LDF option
- [ ] Run ILAMB option
- [ ] Run Crop notebook example

The example for aux_clm might be land_only_testing with fewer variables to process, while ctsm-sci might run land_only with more variables.

Some testing should also be added to CUPiD and cime/CESM itself as well. I think the things that belong there are:

- [ ] Regular CUPiD testing (https://github.com/NCAR/CUPiD/issues/173)
- [ ] cime System test that does CUPiD testing (this might come from the above CTSM SystemTest if we figure this out first)
- [ ] unit test in CUPiD that runs cesm_processing.sh and makes sure it works
- [ ] Tests in CUPiD itself for: LND, ROF, LDF, ILAMB, Crop
- [ ] github actions in CUPiD to make sure the environments are good
- [ ] CUPiD unit tests for the CUPiD python code
- [ ] CUPiD testing for timeaveraging/climo dataset creation / remapping
- [ ] CUPiD testing of each example notebook and plot file creation CUPID_GEN_HTML (seperate command)

- [ ] CUPiD testing that differing number of TASKS give the same answers (does dask promise this?)

(Subtlties of this are dask and non-dask different? Probably. And does it matter for diagnostics?)