Handle Stateful Objects for .topostats

[ns-rse]

NB This issue is mainly an aide-mémoire for @ns-rse and doesn't need to be specifically address in and of itself.

Whilst working through #1102 I've found that I would like to save and subsequently load different stages of processing
to .topostats files so that tests don't require lots of pre-processing (viz. disordered tracing required before
nodestats, nodestats required before ordered tracing, ordered tracing required beofre splining and splining required
before curvature).

Unforutnately because we are adopting/migrating to dataclasses we are unable to pickle such classes as we hit...

            outfile = RESOURCES / "tracing" / "ordered_tracing" / "minicircle_post_nodestats.pkl"
            with outfile.open("wb") as f:
>               pkl.dump(topostats_object, f)
E               TypeError: cannot pickle 'property' object

This is a problem that we will have to solve in some manner because whilst we have methods to convert the classes to
dictionary (see various *_to_dict() methods in topostats.classes) which helps when writing objects to HDF5 we need
to also be able to read HDF5 back into Python and recreate the classes (i.e. TopoStats, GrainCrop Nodes etc.).

There is probably a laborious/manual way of doing this but the "Pythonic" way of doing seems to be to use the duneder
methods __getstate__() and __setstate()__ as detailed in pickle — Python object serialization — Python 3.13.7
documentation (see also
thread).

Its quite possible I'm looking up the wrong avenue for a solution but I think this might be the approach to take so am
noting it here so I don't forget.