Misc packing improvements

[mariosasko]

What does this PR do?

This PR improves the packing logic to make it faster, less error-prone, and easier to read.

The main changes are:

• Replacing the AI-generated BFD splitting (a.k.a. "requeuing") logic from Preserve truncated tokens in BFD packing #4632
  with a vectorized implementation that is significantly shorter and 30% faster.

• Updating pack_examples to restore the input dataset’s format and perform proper input validation.

• Applying a minor optimization to the wrapped packing implementation by reusing the offsets across all packed columns.

• Aligning the naming with the literature (e.g., requeue → split) while preserving backward compatibility.

P.S. The recent Qwen3-Coder-Next technical report includes a nice comparison of packing techniques, which would be a nice addition to the docs. However, the report is not yet available on arXiv, so it cannot be cited as an HF paper.

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?

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.