[Misc] Avoid unnecessary import

[wangxiyuan]

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.

Purpose

Similar with #20900, this PR clean up the unnecessary import (i.e. avoid to import model_executor too early) when vllm setup with online mode. This can fix the usage by vllm-ascend with vllm serve command.

Test Plan

This PR is just for import change. Current test is enough.

Test Result

All test should pass.

(Optional) Documentation Update

[gemini-code-assist]

Code Review

This pull request refactors the codebase to avoid early imports of model_executor, which is a good practice for improving startup time and resolving potential import cycle issues, especially for platform-specific builds like vllm-ascend.

The changes in vllm/entrypoints/openai/speech_to_text.py correctly use a lazy import inside a cached_property to defer the import of get_model_cls.

In vllm/lora/utils.py, the imports from model_executor are moved under a TYPE_CHECKING block, and the corresponding type hints are updated to use string forward references. This is a standard and correct way to handle type-only imports.

The changes are well-implemented and align with the stated goals. I don't see any issues. Good work!

[Yikun]

LGTM, and the reason I also explained in #20900 (review) .

[Yikun]

@DarkLight1337 Would you mind also taking a look?

[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.

🚀