[Feature] Deploy MicroDuck skill policies

[vmoens]

Summary

Adds the MicroDuck-specific integration on top of the generic controller in #4329.

• MicroDuckController maps discrete or parameterized skill decisions to task ids, commands, gait clocks, and low-level actions.
• microduck_skill_env aggregates physical steps and preserves fall summaries for the high-level policy.
• reset_key is public and forwarded to the controller, so environments can use a custom per-agent respawn signal.
• MicroDuckEnv now exposes the real physical fallen signal in its observation spec and outputs.

Review context

This focused PR addresses the two MicroDuck-specific comments from the existing #4329 review while keeping #4329 generic.

Tests

• Native MuJoCo MicroDuck selection: 18 passed, 1 CUDA-only test skipped.
• Controller tests cover discrete and parameterized commands and behavioral reset_key forwarding.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/4392

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit be95118 with merge base 2d258fe ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.