QCSubmit discussion on necessity of InChI and other keys

[lilyminium]

Probably will happen in a QCSubmit meeting.

[j-wags]

We discussed this and I'm fine with the TM datasets going in without InChIs, and really only with the maximal CMILES (AFAIK, nobody on earth has used anything other that the maximal CMILES anyway, and the other identifiers can be derived from it if we need them later). It should be accompanied by a PR to QCADS, but if there's any trouble at all with ingesting the dataset.json, that should be left out and the dataset should be submitted directly using QCPortal (though the PR should go in as a record with the PDF of the structures and such).