Evaluation route choice + renaming references of "congestion" game to route_choice for consistency

[hiazmani]

• Training code for route choice environment
• Renamed the "congestion game" environment (and relevant) files to route_choice for consistency

[ffelten]

I think you need to have a look under doc/, too. There surely are mentions to the old name.

[rradules]

Nice, thanks for handling this. The modifications for the json networks were made to separate the reward signal?

[rradules]

Looks good to me

[hiazmani]

Nice, thanks for handling this. The modifications for the json networks were made to separate the reward signal?

The modifications in the JSON networks are to allow traffic to flow on both directions of each edge. The original graphs allow for agents to use an edge A-B to travel from A to B or from B to A. This doesn't change much in the smaller graphs (Braess) because no route ever uses the edges in the reversed direction, but it does for the larger (city-inspired) graphs. I added the latency function (which allows for computing the rewards) to the backlinks as well to allow this.

Also all the JSON files got "modified" according to Git because the parent directory got renamed from congestion to route_choice.