Remove mps workaround for fp16 GELU, which is now supported natively

[skotapati]

What does this PR do?

Fixes MPS fp16 GELU calls being forced to cast to fp32. fp16 GELU is now supported natively in MPS so this workaround is no longer needed

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?
• Did you read our philosophy doc (important for complex PRs)?
• 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.

[pcuenca]

Thanks a lot @skotapati! Do you know the minimum PyTorch version that benefits from this?

[hlky]

@pcuenca Looks like it's been fixed since 1.13 pytorch/pytorch#86218 the code was originally added #1533 and at that time the minimum supported version for PyTorch was 1.4

[pcuenca]

I tested on 1.13.1 and it still failed, but it worked on 2.0.0. I'd suggest we add a condition to check for a minimum version of PyTorch >= 2.

[skotapati]

Makes sense, adding a check for torch>2.0

[skotapati]

@pcuenca thanks for the review, added the fallback for torch < 2.0

[yiyixuxu]

thanks @skotapati @pcuenca @hlky !