Improve iterative phonon workflow and the output of our phonon workflow

[JaGeo]

To reuse more data from the phonon workflow, the following points have to be improved:

[ ] Optimized structures and isolated atoms energies have to be part of the output!
[ ] add better tests for random structure generation and random seeds! The tests should be all testing the poscar as well.
[ ] remove duplicate test data (I copied some data and only adapted poscars for the new tests in #306 )
[ ] remove duplicated optimization for rattled and phonon structures (there should only be one approach in my opinion)

As a follow up to #306