we changed the way we store agents in Simulation. Now it is a vector of vector of agents, and it seems to perform much better. Thus we are merging all this in v1.0

Many changes, we are almost send a full run. We need now to modify constructor of agent, to distinguish between AgentA and AgentNA.
Then we need to continue unit testing with simulation, maybe adding a callback function.
After that, we are good to go.