Refactor data ingesting in example

[DhairyaLGandhi]

I wanted to reuse some of the examples to scale up training of any property, so that requires some refactoring to clean up data loading etc to not repeat work unless needed.

Quick q: do we know if the cif file reading is thread safe?

[codecov]

Codecov Report

Merging #95 (05946fa) into main (31e3aea) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main      #95   +/-   ##
=======================================
  Coverage   77.38%   77.38%           
=======================================
  Files           2        2           
  Lines          84       84           
=======================================
  Hits           65       65           
  Misses         19       19           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[rkurchin]

I don't know about the thread safety...you'd probably know the better way to test/assess that than I would tbh...

[DhairyaLGandhi]

should help #89 since there were some incorrectly defined paths as well. CI failure seems unrelated, hopefully that doesn't block this PR. We can commit the formatting changes and this should be good. The nice addition would be having a train_formation_energy(graphs) method to share the data ingesting.

[rkurchin]

If this is fine by you and useful for testing you're doing, go ahead and merge (and feel free to bump the patch as well if that's easier). As I crank through the updates to ChemistryFeaturization, I'll be making some other changes (e.g. moving the featurization definition into this package) that will probably motivate a minor version bump, but those are mostly orthogonal to what you're doing here I think.