Post-Step-Hook for External Events

[chenkins]

Objectives
Hock for modifiying state for mimicking external events.

Todos
List what needs to be done for the issue to be considered done and the objectives to be reached.

Minimal testing requirements
Think about what behavior shall be covered by a test (qualitatively describe each test).

Additional context
Relates to #129

Implementation Sketch

• Post-step-Hook just before the rewards (end of episode only) and observations are computed.

        # Check if episode has ended and update rewards and dones
        self.end_of_episode_update(have_all_agents_ended)

        self._update_agent_positions_map()
        if self.record_steps:
            self.record_timestep(action_dict)

        return self._get_observations(), self.rewards_dict, self.dones, self.get_info_dict()

Interface (as long as we have have not a well-defined configuration/state abstraction (probably to be postponed see #127 ):

post_step_hook(env: RailEnv)

Later on:

post_step_hook(i: int, state: RailEnvState)