Clean up previous handling of parallelization

Following discussions, we reverted to an approach where profiles do not have separate parallel_info that can talk to runtime parameters from the calculator, which means the previous code will not be used.

One will therefore still need to do calculator/machine-specific preprocessing in order to set up dynamical parallelization choices. I'm sorry for not having been able to achieve a better solution to this.

Checklist

• I am familiar with ASE's contribution guidelines.
• Doc strings in code changed in this MR are up to date.
• Unit tests have been added for new or changed code.
• Issue is resolved via "closes #XXXX" if applicable.