Make it easier to run the PDDL generator followed by the SAS generator

[FlorianPommerening]

We use this in a lot of examples and is was useful in practice. So far it requires a lot of duplicated code. Different approaches to implementing this could be:

• a utility function to simplify some boiler plate code in the experiment script
• a new pddl_then_sas generator that forwards to the other two generators
• the interview mode generating the scripts as we have them now (including the duplication)

[FlorianPommerening]

Another part of this is the duplication in the evaluators, i.e., that we have to write two evaluators for the PDDL and SAS case that essentially do the same thing. We want to look into sharing code in some way.