JSON QCSchema update

[tovrstra]

Our JSON QCSchema implementation is a bit outdated and incomplete, despite significant efforts to get this to work. (I appreciate all the hard work that has been done.)

The two main problems are:

• Some changes are still pending in PR Add QCSchema Basis #280 that will be difficult to merge due to the large number of PRs that were merged more recently.
• The schema itself has changed recently and can be obtained as mention in Rename json to json_qcschema #360 (comment)

Some points need further discussion before embarking on an update:

• Are there stability guarantees for QCSchema? How is it versioned?
• Validation in our current implementation is done through Python conditionals, which is laborious for both development and maintenance. A more systematic validation with the jsonschema package (or something similar) would simplify this aspect.

[PaulWAyers]

I guess the philosophical issue is whether we should expend effort to support QCSchema, or prioritize it over other json schema associated with quantum chemistry. When I first envisioned including it, I thought it was going to end up being the "universal framework for quantum chemistry calculations" which (I think) was the original intention. But it seems that uptake has been limited; I'm not sure how many packages use QCSchema or QCElemental (which, in my biased view, is basically subsumed by AtomDB anyway). There are a decent number of stars and forks on GitHub, and it is linked into the rest of the MolSSI ecosystem (QCArchive, QCEngine, etc.).

I think that the community has not united around a standard (to my knowledge, though I'm outside the loop here) and perhaps never will given that interconversions become easier and easier nowadays. Philosophically, I tend to believe that it's up to the people who conceive the schema to devise utilities to facillitate their use and interconversion, but that's probably naive (if only because everyone is too busy).

For now, I think I'd welcome any contributions along these lines, and perhaps it is an undergraduate project to implement several of the popular schema at some point. But I feel it's relatively low priority, and can be done at the point that someone using IOData needs to implement it for their own work.