[RLlib] Tuned examples from documentation are not tested

[ArturNiederfahrenhorst]

What happened + What you expected to happen

We recently closed #38968.
The reason the issue was undetected was that we don't test the tuned examples referred to in our documentation.
doc/source/rllib/rllib-algorithms.rst holds a couple of tuned examples today:

Tuned examples:
Unity3D Soccer (multi-agent: Strikers vs Goalie) <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/unity3d-soccer-strikers-vs-goalie-ppo.yaml>,
Humanoid-v1 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/humanoid-ppo-gae.yaml>,
Hopper-v1 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/hopper-ppo.yaml>,
Pendulum-v1 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/pendulum-ppo.yaml>,
PongDeterministic-v4 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/pong-ppo.yaml>,
Walker2d-v1 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/walker2d-ppo.yaml>,
HalfCheetah-v2 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/halfcheetah-ppo.yaml>,
{BeamRider,Breakout,Qbert,SpaceInvaders}NoFrameskip-v4 <https://github.com/ray-project/ray/blob/master/rllib/tuned_examples/ppo/atari-ppo.yaml>

We should at lest execute this for one iteration but even better prove learning for them as a release test.

Versions / Dependencies

master

Reproduction script

none

Issue Severity

None