EAMxx: do not reset yaml file list in eamxx-prod testmod

[bartgol]

Other testmods existing BEFORE eamxx-prod may have already added yaml files.

[BFB]

---

PR #7985 introduced this bug, causing more DIFFs than expected on cdash. I did not catch this, as the eamxx CI does not run eamxx-prod combined with other output-producing testmods.

[mahf708]

Reasonable

[mahf708]

Do we need to start adding them with = instead of +=? If not, I think this should be resolved at the atmchange level, i.e., if a user uses += it should default to start a fresh new list if a list doesn't exist

[bartgol]

There is no such thing as "non-existing" entry (you can't atmchange an entry that doesn't exist). If the array is empty, it simply adds one entry and generate a length-1 array.

[mahf708]

Alright, well, this PR may introduce issues in that this testmod will take all sorts of random files it wasn't meant to take then...

[bartgol]

It only acts on the yaml files introduced by the prod testmod:

output_yaml_files=$(find ${cime_root}/../components/eamxx/cime_config/testdefs/testmods_dirs/eamxx/prod/yaml_outs/ -maxdepth 1 -type f)

Other files are not touched, no?

[bartgol]

Ah, I think what the testmod was doing before the latlon PR was wrong:

        if [ "${file}" == "${output_yaml_files[0]}" ]; then
            # First file, reset output list
            $atmchange -b output_yaml_files="./$(basename ${file})"
        else
            # Append to output list
            $atmchange -b output_yaml_files+="./$(basename ${file})"
        fi  

The testmod assumed that all yaml files in the atm came from eamxx-prod, which is not right. It was basically nuking any existing yamlfile. I am not sure how the eamxx-output-preset-1--eamxx-prod testmods on mappy managed to work. Most likely, the ne4 res causes the 1st file to be skipped, which means the if statement above never triggered.

Anyhow, I think this PR restores things to a more consistent state.