Add ability to perform multiple experiments and logging.

[geektoni]

@AndreiCComan This will enable us to record some specific metrics for each model we want to try. Basically, by using the global logger object, by calling logger.info(something) it will print those line inside an output file, which is named by taking into to consideration the type of EA used and the number of generations (it can be changed though).

Have a look at the evolve() method of the NEATModel to see a possible usage.

[geektoni]

It is possible now to run multiple experiments by calling once evo_flappy.py. For instance, one of the possible procedure call is the following:

python3 evo_flappy.py --MODE_AGENT --MODE_LEARN --EXPERIMENTS --NCPU=4 --MODE_NO_SCREEN --LOG_PERFORMANCE

It will run all experiments, whithout showing the game on screen, and it will log everything inside specific files. NEAT does not work this way, since all its configuration parameter are specified in its config file.

Inside the file experiments.conf all the possible configurations are specified.