Fix Llava conversion for LlavaQwen2ForCausalLM with Clip vision tower

[Isotr0py]

What does this PR do?

Fixes # (issue)

• Some finetuned llava models like FreedomIntelligence/HuatuoGPT-Vision-7B also use LlavaQwen2ForCausalLM architecture, but they use Clip instead of Siglip as vision_tower.
• However, the current llava conversion script hardcoded Qwen with Siglip, which caused the converted model with wrong vision tower config
• This PR adapts this part to make the script more flexible.

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?

@amyeroberts @zucchini-nlp
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]

Hi @Isotr0py, could you show the running, output and sucessful conversion of a previous Qwen checkpoint with these changes?

Otherwise it looks OK to me, but @zucchini-nlp will know which checkpoints we should verify are still compatible with these updates