[Doc][AIR] Improve visibility of Trainer restore and stateful callback restoration

[justinvyu]

Why are these changes needed?

This PR moves the FAQ section on Train experiment restoration to the DL/GBDT user guides. Plus this PR makes these code snippets tested. Secondly, this PR improves the docstrings of methods to implement for saving/restoring stateful Callbacks.

Links for reviewers

• https://anyscale-ray--34350.com.readthedocs.build/en/34350/train/dl_guide.html#fault-tolerance
• https://anyscale-ray--34350.com.readthedocs.build/en/34350/train/gbdt.html#saving-and-loading-xgboost-and-lightgbm-checkpoints
• https://anyscale-ray--34350.com.readthedocs.build/en/34350/train/config_guide.html

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 :(

[matthewdeng]

Very nice!

[matthewdeng]

Probably don't want to link this since this is in the Deep Learning guide.

[justinvyu]

Yeah, I was trying to share some content between the two. I linked it to a section that doesn't really depend on DL trainers. I think it's okay for now, and we can rethink the docs/user guide structure as part of the docs side of the layering project. wdyt?