Remove liger loss in favor of liger kernel

[sergiopaniego]

What does this PR do?

Remove use_liger_loss in favor of use_liger_kernel from trainers and configs.
That flag comes from transformers so we can directly retrieve it from there

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? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

Who can review?

@kashif @qgallouedec

[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.

[kashif]

@qgallouedec we are just investigating the failing tests

[sergiopaniego]

Failing tests seem unrelated

[qgallouedec]

This should be made backward compatible:

• keep use_liger_loss argument, and change its default to None
• in the post init, if a not None value is detected, raise a deprecation warning

  `use_liger_loss` is deprecated and will be removed in version 0.XX. Please use `use_liger_kernel` instead.
  

  and set the value of use_liger_kernel to use_liger_loss

In my opinion, the removal version can be 0.28 in the 3 cases.

[sergiopaniego]

Thanks for the review @qgallouedec! Updated!

[albertvillanova]

Thanks! Just some comments to improve the deprecation cycle for backward compatibility.