Extract labels to a separate metadata tag

The CSV IsotopeData implemention in Euphonic (which will generally be used for applying neutron cross-sections) accepts element symbols with a label following a colon, e.g. `H:2`, `H:D`, `H:my-deuterium` would all be valid and recognised as having Z=1. (A will be determined from the mass and is _hopefully_ 2 for those labels...)

It would be helpful if `abinslib` is also aware of this convention (which is taken from CASTEP) and splits metadata when it is found, i.e. given an input structure with symbols `["H", "H:2"]` the metadata of resulting spectra could have line data `[{"atom_symbol": "H"}, {"atom_symbol": "H", "label": "2"}]`. This would be _very_ convenient for user plotting as they can simply apply a tag to a group of atoms that they wish to plot separately, and then call (possibly indirectly through GUI/CLI) `.group_by("label")` to arrange the results on this basis.

*Duplicated from https://github.com/pace-neutrons/abinslib/issues/35*

Start by tracing how abinslib converts input structure symbols into spectra metadata; the issue names no specific file or test. Verify handling of symbols such as H:2, H:D, and H:my-deuterium, and confirm that resulting metadata separates atom_symbol from label so grouping by label can distinguish tagged atoms.

Written by the indexing model from the issue text.