`Workflow.analyze_structures` has no consistent return behaviour

Depending on the workflow you might get a dict, or a tuple of floats, or a tuple of dicts. In each individual case these make perfect sense, but overall the variance makes the Workflow class feel disjointed. What about enforcing consistency across return values, e.g. Workflow.analyse_structures(self, output_dict: dict) -> dict?

Advantages:

• Workflow return values seem pretty clear in the code, e.g. return eng_pot, kinetic_energy, but this is lost to the user who just sees two floats and needs to go look at the docs to figure out their meaning; with a dict those nice variable names could be directly visible
• It makes it easier to consistently leverage the workflows downstream; e.g. in pyiron_workflow we have a node to extract workflow results and it's named result_dict, but this is not true for some of the workflows even though the node is working perfectly fine.

Ideally one could even use something like a DotDict so the results are both named and tab completable.

Start by comparing the return paths of Workflow.analyze_structures across the workflow implementations, including atomistics/workflows/langevin/workflow.py around line 137. Review the linked pyiron_workflow DotDict and result_dict usage to define a consistent result contract. Done means workflows expose consistently named results and downstream extraction remains compatible.