[Feature] Train MicroDuck skills and navigation

[vmoens]

Summary

• add the focused MicroDuck PPO training launcher
• support task remapping and optional critic restoration when loading checkpoints
• cover checkpoint restoration across navigation, skill, and critic parameters
• exclude the long-running launcher from generic example execution and smoke-test it in the native MuJoCo tutorial environment

This PR is intentionally limited to the training entry point. Its temporary synthetic base collects the focused dependencies from #4329, #4388, #4389, #4390, #4391, #4392, #4393, #4394, #4395, #4396, and #4398. It will be rebased onto main as those land.

Test plan

• launcher checkpoint/load/remap tests: 3 passed
• full two-stage launcher smoke run: passed
• python -m examples.microduck.train_skills --help: passed

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/4397

• 📄 Preview Python docs built from this PR

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

❌ 2 New Failures, 1 Unclassified Failure

As of commit 24b144f with merge base 2d258fe ():

NEW FAILURES - The following jobs have failed:

• Unit-tests on Linux / tests-cpu (3.10, bulk) / linux-job (gh)
  test/libs/test_mujoco.py::TestMujoco::test_microduck_training_launcher_restores_walker[True-False]
• Unit-tests on Linux / tests-cpu (3.13, bulk) / linux-job (gh)
  test/libs/test_mujoco.py::TestMujoco::test_microduck_training_launcher_restores_walker[True-False]

UNCLASSIFIED FAILURE - DrCI could not classify the following job because the workflow did not run on the merge base. The failure may be pre-existing on trunk or introduced by this PR:

• MuJoCo Custom Envs Tests on Linux / unittests-mujoco (3.11, 12.8) / linux-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  test/libs/test_mujoco.py::TestMujoco::test_microduck_training_launcher_restores_walker[True-False]

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