Item gathering added to main

[rradules]

The implementation of the MO-Item gathering environment based on the environment from:

Johan Källström and Fredrik Heintz. 2019. Tunable Dynamics in Agent-Based Simulation using Multi-Objective
Reinforcement Learning. Presented at the Adaptive and Learning Agents Workshop at AAMAS 2019.
https://liu.diva-portal.org/smash/record.jsf?pid=diva2%3A1362933&dswid=9018

[ffelten]

Good one :-).
My main worry is essentially the divergence with the original env. We can discuss that later today.

[ffelten]

Cleanup prints :-)

[rradules]

Thanks, I made all the updates. Biggest change is adding functionality to generate random maps and randomise maps, according to the original paper. I also added additional notes on how our observation space is different than the original implementation.