Main errors right now appear to be with the address sanitizer, which is to be expected with these many changes. Yep, some mistakes on my part.

Hmm, now the pytests are failing 🤔 I would have thought that these would pass if the ctests were passing.

Hmm, now the pytests are failing 🤔 I would have thought that these would pass if the ctests were passing.

There are many naively distorted crystal structures with respect to ideal symmetry. ctests covers very basis ones, but pytest does more cases. It results in entering specific routines to recover from detailed different symptoms.

I feel this PR is dangerous. It would be helpful to divide into small commits.

I feel this PR is dangerous. It would be helpful to divide into small commits.

Indeed that would be ideal, but it will take a bit of time yo detangle them. At least the safe ones we have it in the other 2 commits.

There are many naively distorted crystal structures with respect to ideal symmetry. ctests covers very basis ones, but pytest does more cases. It results in entering specific routines to recover from detailed different symptoms.

Hmm I think we should address #291 and its equivalent googletest to actually report the specifics of how it fails. It should be possible to map the pytests 1-to-1 onto the ctests, but it would require a bit of organization, mainly split the unit-tests from the functional tests. All of the python and fortran api tests are only functional ones, and those should be identically covered and use a common set of inputs that we synchronize through a file. For this testpytests should be moved to /tests folder which we then split logically as well.