[Misc] Collect model support info in a single process per model

[DarkLight1337]

Follow-up to #9208

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
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 do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀

[youkaichao]

LGTM

[youkaichao]

please fix the error. it looks oot registered models are not processed correctly.

[DarkLight1337]

It looks like changes to the registry (via register_model) aren't being preserved in other processes. Is there a good way to solve this?

[DarkLight1337]

I'm thinking of loading the plugins for each process.

[DarkLight1337]

I have fixed this by adding the load_general_plugins call to MQLLMEngine.from_engine_args as well.

[youkaichao]

It looks like changes to the registry (via register_model) aren't being preserved in other processes. Is there a good way to solve this?

good point. the new process does not load plugins. maybe we can just add one line of loading plugin in the if __name__ == "__main__":

[tlrmchlsmth]

Really timely PR, as I was about to make this worse in #6484 😀