Fix Transformers backend tensor parallel for multimodal models

[hmellor]

This PR fixes 2 things:

• Top level multimodal configs don't have base_model_tp_plan. So we make sure we get it from each sub config in TransformersBase
• If a model with a TP plan is a sub model, the to plan patterns won't match because there is a missing prefix. We now add this prefix before matching.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[gemini-code-assist]

Code Review

This pull request introduces two targeted and important fixes for enabling tensor parallelism in multimodal models using the Transformers backend. The first change correctly retrieves the tensor parallelism plan from the text model's configuration, which is the appropriate location for multimodal architectures. The second change smartly restricts the application of tensor parallelization to only the language model component, which correctly prevents unintended and problematic modifications to the vision tower. Both changes are well-implemented and appear correct. I have no further comments.