Replace sub-processed ppfp usage with the pp-format python library for extract validation

[mo-jareddrayton]

There's a significant overhead introduced by subprocessing out to the ppfp command. We can rewrite the get_stash_from_pp function using ppformat directly with something like the following.

from pp_format.pp_format import list_fields

def get_stash_from_pp(filepath) -> dict[str, int]:
    fields = list_fields(filepath, ["STASH"])
    stash_list = [str(field_dict["STASH"]) for field_dict in fields]
    return Counter(stash_list)

The tricky bit is how we manage the dependency as pp-format is on neither pypi or conda.

[matthew-mizielinski]

We would like to take pp format public -- see https://github.com/MetOffice/pp-format/issues/2