Refactor inner training loop

[winglian]

What does this PR do?

The _inner_training_loop method has a lot going on which makes it hard to extend for downstream developers/libraries. This PR breaks it up into smaller well described methods that are chained in the training loop.

Additionally, this adds the concept of a DataProducer that can wrap the dataset/sampler/dataloader which allows for better handling of Iterable data and Async data for use cases like async RL downstream (e.g. trl)

Fixes # (issue)

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: metaclip_2