Fix image-text-to-text stop_sequence handling

[Sunt-ing]

What does this PR do?

ImageTextToTextPipeline accepts stop_sequence, but the stop-sequence branch writes into generate_kwargs even when the user did not pass a generate_kwargs dict. A normal pipeline call such as pipe(image, text=..., stop_sequence=".") fails before generation with:

TypeError: 'NoneType' object does not support item assignment

This creates the forward generate_kwargs when needed and passes stop sequences through generation's stop_strings support, matching the any-to-any pipeline path. This also avoids forcing an extra eos_token_id into image-text model wrappers that already provide their own decoder EOS value.

Repro and output

from PIL import Image
from transformers import pipeline


model_id = "Xenova/tiny-random-LlavaForConditionalGeneration"
image = Image.new("RGB", (16, 16), color="white")
pipe = pipeline("image-text-to-text", model=model_id, device=-1)
print(pipe(image, text="<image> Describe.", stop_sequence=".", max_new_tokens=1))

# current main
loading Xenova/tiny-random-LlavaForConditionalGeneration
loaded ImageTextToTextPipeline LlavaForConditionalGeneration
ERR TypeError "'NoneType' object does not support item assignment"

# after this PR
loading Xenova/tiny-random-LlavaForConditionalGeneration
loaded ImageTextToTextPipeline LlavaForConditionalGeneration
OK [{'input_text': '<image> Describe.', 'generated_text': '<image> Describe. starb'}]

Tests

python -m pytest -q tests/pipelines/test_pipelines_image_text_to_text.py::ImageTextToTextPipelineTests::test_stop_sequence_without_generate_kwargs

tests/pipelines/test_pipelines_image_text_to_text.py::ImageTextToTextPipelineTests::test_stop_sequence_without_generate_kwargs PASSED [100%]

============================== 1 passed in 0.79s ===============================

python -m ruff check src/transformers/pipelines/image_text_to_text.py tests/pipelines/test_pipelines_image_text_to_text.py
All checks passed!

git diff --check

Code Agent Policy

• I confirm that this is not a pure code agent PR.

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 and the
  Pull Request checks?
• 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 according to the guidelines?
• Did you write any new necessary tests?

Who can review?

cc @zucchini-nlp @Rocketknight1

[github-actions]

---

CI recap

Dashboard: View test results in Grafana
Latest run: 28625110499:2
Result: success | Jobs: 15 | Tests: 63,630 | Failures: 0 | Duration: 17h 35m