Pipeline VQA: Add support for list of images and questions as pipeline input

[BlacCod]

What does this PR do?

Fixes #31216

This PR will allow passing list of either PIL.Image or str as pipeline argument (and list[str] for questions). Passing dataset (in form of .pt_utils.KeyDataset) is also allowed.
So this code will now work (passing the dataset as list)

from transformers import pipeline
dataset = datasets.load_dataset("frgfm/imagenette", "full_size", "validation[:10]")
oracle = pipeline(task="vqa", model="dandelin/vilt-b32-finetuned-vqa")
oracle(question="What's in this image?", image=dataset['image'], top_k=1)

This also works:

from transformers.pipelines.pt_utls import KeyDataset

oracle(question="What's in this image?", image=KeyDataset(dataset, "image"), top_k=1)

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.

[amyeroberts]

Thanks for adding this capability!

Seems like something which would be useful to have. Alongside the logic, there's a few things we should add alongside:

• Update docstrings to reflect the new input types
• Add tests for the new input types

[BlacCod]

Hi @amyeroberts, thank you for reviewing :) I added the tests and docstrings
As for the docstring, I assume you mean the comments inside the function? There isn't much to update there (the original author wrote the "correct" documentation, but did not implement everything for some reason).

[BlacCod]

Found and fixed the problem with the dataset :) Now dataset (in form of KeyDataset, to stay consistent with this tutorial) as input for image is accepted

[amyeroberts]

Great - thanks for adding this!

[amyeroberts]

Thanks again for this contribution!