[WIP] Add error route for molecules in optimize export

[j-wags]

Description

This PR is an alternative solution to #58. I think it's safer, since it covers a wider range of possible issues. However I'm a bit concerned because I can't test it.

Todos

• [ ] Add tests
  • [ ] A test for successful optimization could take a pre-optimized structure, which could complete in 1 iteration
  • [ ] A test for the stereochemistry failure I suspect we're seeing could take a N=N double bond, twist it to around 90 degrees, submit it, then immediately try to export the input conformer.

[dotsdl]

Thanks for putting this together @j-wags; I like your solution, and will review this tomorrow and work to address the concerns you have raised.

[j-wags]

I'm going to close this PR, in favor of consolidating the error handling logic in the PR where we handle proton transfer.