What does this PR do?

1. Improve fps handling in Qwen2_5_VLProcessor.__call__()

• Previously, the __call__() function always defaulted to fps=2, regardless of the input video.

• Now, when fps is not explicitly provided and cannot be read from the video metadata, it falls back to 24 by default. Otherwise, the provided or metadata fps is used. It follows a priority order:

  kwargs["fps"] → video_metadata["fps"] → 24 (default)

  The default value 24 is chosen as it represents a common frame rate for most videos.

• This prevents potential errors in second_per_grid_ts calculation when the actual video frame rate differs from the assumed default.

2. More robust do_sample_frames behavior

• If fps is manually provided but do_sample_frames is not specified, do_sample_frames will now be automatically set to True.
• This avoids cases where custom fps is ignored due to missing sampling flag.

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?

@zucchini-nlp