Fix for QCFractal next beta10

[loriab]

Description

closes #2939

User API & Changelog headlines

• fix bug where optimize(..., engine="geometric", molecule=mol) would not run
• undo some hack fixes and make distributed driver runs through QCFractal "next" branch more reliable by making array qcvars shaped numpy arrays. Works with QCFractal >=v0.50beta10. Show examples of using get_psi_results() to prepare the results of distributed driver runs in "normal" psi4 way.

Dev notes & details

• In the early days (before ddd entered psi4), all array-like pieces of AtomicResults came back as flat lists, so we had to "plump" them into e.g., (nat, 3) for gradient. Then, by the time of v1.6 and QCFractal master v15.8, the AtomicResults preserved all that, so we stopped using the plumper. Now in qcfractal next, they're back to lists again and qcvars have been mingled with other properties, so both separation and plumping are required.
• promote get_psi_results as a drop-in replacement after running distributed. It sets up a dummy wfn and qcvariables, just like an internal run, so one doesn't have to deal with picking results off the AtomicResult QCSchema.
• Added labels to tests (mostly findif) so that one can catch all ddd-using tests with pytest -m "findif and cbs and nbody"
• set up a mechanism where can parameterize both pytests and ctests with a snowflake with minimum clutter.

Questions

• This won't work with current qcf release. I'll ask @bennybp to mint a beta10, then this should work.

Checklist

• Tests added for any new features
• All or relevant fraction of full tests run

Status

• Ready for review (for review, the 4th commit has the material changes. 3rd are test changes. 1st & 2nd are labeling and syntax.)
• Ready for merge

[maxscheurer]

Is pprint used or just a debug leftover?

[loriab]

It's a debug leftover that I keep adding again and again, so I thought I'd leave it (until some linter confiscates it).