Generate Union of Causal Net Predictions

[dbdimitrov]

Some MILP problems result in multiple equally good (local) solutions. A non-arbitrary way (independent of random state) to deal with those is to create a union of them.

SCIPy will additionally not converge properly with larger problems (now more notable due to recent update and size increase of the OmniPath PPI).

• Implement an option to internally generate a union of solutions
• Change example to use Gurobi