Reuse network topologies between sub-experiments

[castarco]

Right now, every sub-experiment generates a new random preferential attachment graph (using the Barabási–Albert model) to define its network topology.

In order to make comparisons between different settings sets more fair and useful, the sub-experiments should be executed using the same network topology.

Subtasks:

• Ensure that all sub-experiments share network topology.
• Enhance the experiment runner by allowing to execute the same sub-experiment many times on different network topologies (still shared with the other sub-experiments).