Add API / wrappers for life long learning RL

[pseudo-rnd-thoughts]

In discussions with a PhD college, they were explaining that life long learning in RL doesn't have a set of environments, API, wrappers, etc for researchers to use.
Therefore, I would propose that we add support for this as it is related to meta and multi-task RL.
In addition, we could look at transfer learning as an additional type of RL which I don't know the status of.