Function train_eval_doom_simple

A simple train and eval for PPO.

(
		# Params for collect
		num_environment_steps=30000000,
		collect_episodes_per_iteration=32,
		num_parallel_environments=32,
		replay_buffer_capacity=301,  # Per-environment
		# Params for train
		num_epochs=25,
		learning_rate=4e-4,
		# Params for eval
		eval_interval=500,
		num_video_episodes=10,
		# Params for summaries and logging
		log_interval=50)