[RLlib] Add Humanoid + SAC tuned_example script.

[sven1977]

Add Humanoid + SAC tuned_example script.

Not fully tuned yet, but this setup already learns up to 1200 return (in 10h on single GPU with SAC). Probably just needs to run longer for even better results.

Just wanted to lock this in already as a milestone.

Why are these changes needed?

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
  • I've added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[simonsays1980]

LGTM. Please check the learning rates and my comment.

[simonsays1980]

I wonder how this can run with a learning rate not None? The new PR is merged and requires lr=None in the new stack. Instead actor_lr, critic_lr, and alpha_lr can be used.

[sven1977]

True, let me double check ...

[sven1977]

Great catch! I must have still run it on the old master (before we merged that PR).

Fixed.