feat: add BLIP support in TransformersImageToText

[anakin87]

In Hugging Face Transformers, support for BLIP in the image-to-text pipeline has recently been added (huggingface/transformers#21904).

BLIP is a powerful model, so this PR is intended to bring this improvement to Haystack.

Proposed Changes:

• extend the list of supported models, including BLIP
  (this list is still needed because some models that can be loaded via HF pipeline raise errors)
• make Salesforce/blip-image-captioning-base the default model: it is way better than nlpconnect/vit-gpt2-image-captioning and is about the same size
• Remove a test that is now irreproducible in the sense that it originally was (due to changes in implementations in HF transformers)

How did you test it?

Existing tests in CI.

Other related issues

• Add support for BLIP and GIT in image-to-text and VQA pipelines huggingface/transformers#21110
• Impossible to use any of the GIT models from Microsoft #4301
• docs: TransformersImageToText- inform about supported models, better exception handling #4310

Checklist

• I have read the contributors guidelines and the code of conduct
• I have updated the related issue with new insights and changes
• I added tests that demonstrate the correct behavior of the change
• I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test:.
• I documented my code
• I ran pre-commit hooks and fixed any issue

[coveralls]

Coverage: 37.893% (+0.1%) from 37.792% when pulling 150fe00 on anakin87:blip_support into 6eb251d on deepset-ai:main.

[silvanocerza]

All good, just a small nitpick in the docstring.

[anakin87]

@silvanocerza feel free to merge 😊