fix atom_types parsing

[lukaspie]

This fixes the additional issue described in #583 (comment). The problem was that

(atom_types := sample.get("atom_types__field"))

does not actually return the value of sample.atom_types__field, but just the string atom_types__field. So we need to use the workaround presented here.

Local script for checking (using nomad==1.4 and pynxtools from this branch):

from nomad.datamodel import EntryArchive
from nomad.utils import get_logger
from pynxtools.nomad.parser import NexusParser

archive = EntryArchive()
NexusParser().parse(str("output.nxs"), archive, get_logger(__name__))

assert archive.data.ENTRY[0].specimen.atom_types__field == "C, Cr, Cu, O, Si"
assert archive.results.material.elements == ["C", "Cr", "Cu", "O", "Si"]

Here's the result section when running the GUI and appworker locally using the versions above:

@mkuehbach please test locally.

[lukaspie]

We unfortunately do not have a test for this at the moment (deactivated at https://github.com/FAIRmat-NFDI/pynxtools/blob/master/tests/nomad/test_parsing.py#L222) as atom_types is not part of NXsample and we do not have a NeXus file in the repo that uses an application definition where atom_types is defined. Could potentially add an APM/MPES file where this is used for testing.

[mkuehbach]

Fix makes sense.

(in the original version atom_types was always instantiated irrespective whether the if branch is taken --- eventually to such that atom_types ends up as None)