[WIP] Update VASP outputs.py for more robust piezo and elastic tensor extraction from OUTCAR

[knoori]

Purpose:
This PR updates the VASP outputs.py to allow for the extraction from OUTCAR of piezoelectric properties from PEAD calculations (LCALCEPS). It also updates the code to allow for the correct extraction of the electronic (ion-clamped) piezoelectric stress tensor for both LEPSILON and LCALCEPS calculations. The full elastic tensor is now also extracted in the case of a finite difference (FD) calculation that allows for variable cell deformation.

Modifications:
The Outcar class is modified to include a finite difference self.fd check if IBRION is 5 or 6. We also check for RPA (LRPA) in OUTCAR and whether a variable cell relaxation is being performed (ISIF is 3), and set self.lrpa and self.varcell accordingly.

We add a check for LCALCEPS in OUTCAR and set self.lcalceps accordingly. Similar to the workflow for LEPSILON, we call self.read_epsilon() if self.lcalceps is True, and call self.read_lepsilon_ionic() if self.fd is True. Note that we can re-use the read_epsilon() and read_lepsilon_ionic() functions for the LCALCEPS workflow.

We update read_lepsilon() to check if self.lrpa is False (it should be for piezo calculations, cf. VASP wiki) and modify the search string to correctly extract the electronic piezo tensor (verified to work at least for VASP v6.4.3).

If we have a FD calculation (self.fd = True) and ISIF is 3 (self.varcell = True), we can extract the full elastic tensor using self.read_elastic_tensor(). This function has been updated by modifying the search string to ensure the full 6x6 tensor is correctly read (verified for VASP v6.4.3).

In as_dict we add checks for self.lcalceps, self.lcalceps AND self.fd, and self.fd AND self.varcell. We append the corresponding electronic piezo, ionic piezo, and elastic tensors in a manner consistent with the existing code.

NB I’m not sure if the changes are substantial enough to warrant new tests, but I can write them if needed.