Closing. This seems very unspecific and/or already done. Whatever we wanted to do (if we were clear about it) also very likely no longer fits the changes incurred by our proposed solution for the component interaction problem.

In issue774 we introduce a PerTaskInformation. Here we want to develop a better
way of how use it to handle task data. One question is how to support
polymorphic behavior, e.g., how can a CostAdaptedTask re-use the causal graph of
its parent.