Add Q-learning and SARSA agents

[riiswa]

Description

This PR add 2 tabular agents: Q-Learning and SARSA (continuity of #254).

TODO

Add 1 example for qlearning and 1 for sarsa. (video plot)

[KohlerHECTOR]

Hello it looks good to me, do we usually use the typing library ?

[riiswa]

Hello it looks good to me, do we usually use the typing library ?

Yes (see #293)

[TimotheeMathieu]

Looks good to me. I did not check the code carefully but it seems ok.
The docstrings are nice, maybe you can add a "attributes" section (with self.Q typically to say where to go search for Q function) and an examples section (those are optional for now in rlberry but it is nice to have them for QL and SARSA because these are basic algo). See numpy docsting for more info on how to style these sections.

Otherwise, you should also add the two agents in api.rst for them to appear in the documentation, add the change to the changelog in changelog.rst and add the agents to the regular tests in test_agent.py and yes as you pointed out examples would be nice :)

Thanks!

[riiswa]

Done, sorry for the delay @TimotheeMathieu :)