Exchange solvers

[aiAdrian]

Objectives

Different solvers (e.g. Policy) should be testable without changing the general Flatland.

• A general Flatland Problem Solver framework should be implemented

Todos

What needs to be done for the issue to be done and the objectives to be reached:

• [x] Flatland Problem Solver Generalisation has to be implemented
• [x] Exchangebale learning Multi-Agent-Reinforcement-Policy and optimisation strategy must be implemement, such as DDDQN, PPO, ...
• [x] Heuristic solvers (policy) have to be implemented to demostrate that Flatland is more than MARL

Minimal testing requirements

Policy exchange capability can be demostrated

• [x] A working examples is available

[aiAdrian]

I did a first implementation: https://github.com/aiAdrian/flatland_solver_policy