This is a selection of spectral lines from splatalogue for
distribution with CASA.  Markus got it from
https://safe.nrao.edu/wiki/bin/view/ALMA/CASA_Offline_Splat_list
mainly for LineTAP experiments.  Some documentation is also findable
through https://splatalogue.online/.

The one big challenge here are the inchis, as whatever the NRAO folks
have as identifiers is a crazy hodgepodge that in part doesn't even seem
to make chemical sense (like stereo markers (?) on non-chiral molecules)
and either way cannot reliably be parsed by a machine.

To come up with halfway reliable InChIs anyway, I'm extracting the
"Species" and "Chemical Name" columns and sort them such that similar
things hopefully are next to each other; that's the root of the
res/species-key.txt file and is implemented in bin/extract_species.py.

Next, I manually annotate that file with a third column containing the
structure formula in our heuristic protocols.linetap code.

The result of that is then processed with bin/compute_inchis.py, which
add the inchis to res/species_key.txt.  That is then used by the
importer.

If you do big changes here, see species/README on how to propagate
the fixes here to there.

The test set is the first 1000 lines of the TSV file.