[masking] fix Aggressive boolean conversion breaking packing implementations

[kashif]

What does this PR do?

Fix Aggressive boolean conversion in the masking_utils that is breaking packing implementation

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.

[Copilot]

Pull Request Overview

This PR fixes an aggressive boolean conversion issue in the masking utilities that was breaking packing implementations. The changes include:

• Adding a regression test to ensure non-binary (fractional) attention masks are preserved.
• Updating _preprocess_mask_arguments to detect binary masks via unique values and preserve non-binary masks.
• Adjusting flash_attention_mask to correctly handle boolean and non-boolean masks.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
tests/utils/test_modeling_utils.py	Adds tests for preserving non-binary attention masks and ensuring correct flash attention behavior.
src/transformers/masking_utils.py	Updates boolean and non-boolean mask processing in _preprocess_mask_arguments and flash_attention_mask.

[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.

[Cyrilvallez]

Superseded by #39194! Closing!