Trainer Refactor: Part 1

[muellerzr]

What does this PR do?

This is the first chunk in my attempt to clean up the trainer without breaking core capabilities (and integration capabilities/projects).

The aim is for readability rather than enhancement of features. I've started this process by:

• Refactoring how we get the "core" training values (max_steps, update_steps_per_epoch, num_train_tokens) in a more flow-based pattern than before (e.g. grouping num_train_tokens together)
• All *_steps based setups for the Trainer.state are now a func inside of TrainerState (a user should only need to know that's being done, realistically)
• Similarly there is a init_training_references which transparently hints that things are being stored in the state
• I've also refactored the RNG state setters to a helper func, to reduce repeated code that will repeat as more device support comes.
• Lastly I've simplified a bit of the accelerate-based setup items.

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.

@SunMarc
@tomaarsen
@Rocketknight1

[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.