Idea: Optimize markov legends/vectors

[pvdz]

The markov vars are relatively quite expensive as their probability table is built up on every step of the way. We should take a look at this and see if we can pre compute certain parts of this table. And for example prune stuff that is never going to be used. The smaller the legend of a markov var the faster it will run.

• May not affect speed that much
• May be complex with the implicit vector expansion
• Legend elements whose probability vector is 0 should be eliminated entirely. Their domains should be pruned from these options
• There could be multiple steps which would make this idea very hard, if not impossible, to implement. Take care about the various edge cases of markov configs.