Improve logging and multi-processing.

[jangroter]

The current multi-processing script has two bugs:

Bug 1 and proposed solution
When initializing the seperate environment instances for the different CPUs each instance uses the same seed. This is solved by using Stable-Baselines3's own SubprocVecEnv class, which handles this naturally.

Bug 2 and proposed solution
The current logger implementation only logs the first environment instance when using multi-processing, the proposed change loops through all the environments done flags and adds them to their own row when necessary. The new log-files contain an additional env_id flag.

Example to demonstrate logging functionality over multiple parallel instances:

timesteps,episodes,env_id,total_reward,total_intrusions,average_drift
81,1,2,-3.268373764085395,0,1.2105088015131091
252,2,0,-34.84157606630309,26,1.0525685793217985
291,3,2,-31.89786281975591,27,0.6996946885365583
351,4,1,-33.25546853487047,26,0.6201255158008938
408,5,0,-76.65059943658756,69,1.4712691224206815

[jangroter]

TODO before merge:

• Check compatibility of logger with all environments (both multi-processing and single core)