In translate.py, we create two var/val encodings of facts, one for the FDR representation we 
use at the end and one for an intermediate representation f the full mutex groups (cf. the two 
calls of strips_to_sas_dictionary). We build a mutex_key to get from one encoding to the other. 
We should check whether we could not use the same encoding for both from the beginning. From a 
quick check, I did not even see where argument mutex_dict of translate_task is ever used.

Currently, we also have a duplicate type alias declaration for VarValPair. Once there is only a 
single encoding, we should everywhere use the one from sas_tasks.py. In the current situation I 
wanted to avoid this, because it would give the wrong impression that such a Pair always refers 
to the final encoding.