Non blocking support to torch DL's

[muellerzr]

What does this PR do?

This PR brings in huggingface/accelerate#2685 to transformers.

TL;DR Accelerate's dataloaders can optimize send_to_device when sending the data to CUDA directly using non_blocking safely. This PR allows the user to configure that as part of the AcceleratorConfig via non_blocking=True.

It's also recommended to use pin_memory=True, so this will flag a warning letting the user know it's recommended if not enabled.

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.

@amyeroberts

[amyeroberts]

LGTM!

[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.