Tammo van der Heide

Thanks for splitting the PR, I'll check the forces against [calcderivs](https://github.com/dftbplus/dftbplus/blob/main/utils/numcheck/calcderivs) (probably next week).

@bhourahine @thomas-niehaus Seems to improve agreement of analytical/numerical forces, at least for my test cases, by about one order of magnitude, i.e. ~5e-03a.u. (#947) vs. ~3e-04a.u. (#950). --> Test system:...

Hi Ben, I just stumbled across this PR. Can you already foresee if the way of MPI parallelization as you want to implement it here also works for the [gamma-point...

Buildbot tests still running, but since the error I've encountered in #1117 was not noticed before this is probably pointless.

@conda-forge-admin, please ping team

@conda-forge-admin, please ping team

@conda-forge-admin, please ping team

@jubich Could you resolve the conflict with main?

@carterbox I have incorporated all the suggestions of the code review and resolved the overdependencies. Could you take another look at it?

@bhourahine I did not yet find time to look into it again, will try to do so soon.