Miscellaneous environments

I would like to also implement a set of classic miscellaneous environments, which are not part of the standard environment APIs. Here is a list to address:

[x] Different Bernoulli Bandits (as in Wang et al., RL^2).
[x] Gaussian Bandit as in (Lange & Sprekeler, 2020)
[x] Sutton's Four Rooms environment.
[ ] Markow reward processes (as in MPG work).
[ ] Markov chain processes (as in MPG work).

RobertTLange / gymnax

Miscellaneous environments #9