Use Dataclass for ASE and QE

[jan-janssen]

No description provided.

[samwaseda]

From my point of view, it makes more sense to have pre-defined fields in AtomisticsOutput, such as energy, forces etc., since we have already seen in pyiron_atomistics that by not defining fields, we could easily end up having different tags for the same values (cf. SPHInX & VASP)

[jan-janssen]

From my point of view, it makes more sense to have pre-defined fields in AtomisticsOutput, such as energy, forces etc., since we have already seen in pyiron_atomistics that by not defining fields, we could easily end up having different tags for the same values (cf. SPHInX & VASP)

Done

[samwaseda]

Here I see the same problem as the one we discussed here and here. To the very least, these lines appear twice inside atomistics, so they have to be somehow refactored, but more fundamentally, I see a clear connection between tags, tasks (i.e. calc_XYZ) and interactive getters. Interactive getters and the tags are now fully conceptualised with the data class, and I think the same should happen for the tasks as well.

[liamhuber]

110%, although fair to break that step into a different PR

[liamhuber]

It's not mission critical because one can always refactor: rename later, but it might be wise to resolve #129 before/while propagating these classes to the other calculators

[samwaseda]

There's still no concept for the part that I mentioned, but it's now refactored, so I'm ok with the current state

[jan-janssen]

There's still no concept for the part that I mentioned, but it's now refactored, so I'm ok with the current state

I agree I played around with it a bit and was not able to solve it, so I would prefer to move forward with the current pull request and then do the optimisation in a separate pull request.