Amset interpolation convergence

[Jackbt21]

copy_amset_files() currently looks for a file named transport.json. AMSET writes the transport files as transport_{mesh}.json by default. Because that file is never found, transport.prev.json is never created, so check_converged() has nothing to compare against on a resubmission, leaving jobs resubmitting weather the interpolation factor is converged or not.

Summary

• Replaces the exact match lookup of transport.json with the wildcard pattern transport_*.json*, matching the default file naming.
• rename_files() doesn't support wildcards, so the transport file is renamed with a plain pathlib glob + rename instead of going through rename_files().
• Added allow_missing=True into copy_files, replacing what was in rename_files, to avoid crashing on the first run when no previous file exists.

[JaGeo]

@utf can you please check?

[esoteric-ephemera]

Would a safer wildcard pattern be "transport*.json" so the previous matches would still be there (if applicable)?

[Jackbt21]

In any of the later iterations (from the 3rd onwards). The previous directory will contain transport.prev.json* and transport_{mesh}.json*. The underscore is here to differentiate between the two so the most recent transport file is selected and not the old one.

[esoteric-ephemera]

Change to this? Calling Path with a regex interprets the symbol literally, AFAIK

files.extend(list(Path(".").glob("transport*.json")))

[Jackbt21]

This change was introduced to satisfy the mypy type checking in the test suite, which expects files to contain either Path or None. The path stored inside the Path object is later treated as a globbable string, rather than being subject to any of the class methods, so the Path is essentially just a wrapper.

Alternatively, I found that the mypy type checking can be satisfied by adding a type hint and appending "transport_*.json*" as a string:

    # find optional files
    files: list[str | Path] = []
    for file in (
        "settings.yaml",
        "vasprun.xml",
        "band_structure_data.json",
        "wavefunction.h5",
        "deformation.h5",
    ):
        found_file = get_zfile(directory_listing, file, allow_missing=True)
        if found_file is not None:
            files.append(found_file)

    files.append("transport_*.json*")

Either way, the wildcards are handled later in common.files.find_and_filter_files, as called by copy_files and gunzip_files.

I am happy to go with either solution.