[Refactor] Lazy import tool_parser

[chaunceyjiang]

Purpose

Lazy-loaded tool_parser

Fix https://vllm-dev.slack.com/archives/C0911AKUZQX/p1762156010866119

Test Plan

vllm serve --help=all
...
                        The file path to the SSL key file. (default: None)
  --tool-call-parser {deepseek_v3,deepseek_v31,ernie45,glm45,granite,granite-20b-fc,hermes,hunyuan_a13b,internlm,jamba,kimi_k2,llama3_json,llama4_json,llama4_pythonic,longcat,minimax,minimax_m2,mistral,olmo3,openai,phi4_mini_json,pythonic,qwen3_coder,qwen3_xml,seed_oss,step3,xlam} or name registered in --tool-parser-plugin
                        Select the tool call parser depending on the model that you're using. This is used to parse the model-generated tool call into OpenAI API
                        format. Required for `--enable-auto-tool-choice`. You can choose any option from the built-in parsers or register a plugin via `--tool-
                        parser-plugin`. (default: None)
...

vllm serve /home/jovyan/qwen3-8b --reasoning-parser qwen3 --guided-decoding-backend xgrammar --enable-auto-tool-choice --tool-call-parser hermes
...
APIServer pid=145800) INFO 11-03 10:06:56 [launcher.py:46] Route: /scale_elastic_ep, Methods: POST
(APIServer pid=145800) INFO 11-03 10:06:56 [launcher.py:46] Route: /is_scaling_elastic_ep, Methods: POST
(APIServer pid=145800) INFO 11-03 10:06:56 [launcher.py:46] Route: /invocations, Methods: POST
(APIServer pid=145800) INFO 11-03 10:06:56 [launcher.py:46] Route: /metrics, Methods: GET
(APIServer pid=145800) INFO:     Started server process [145800]
(APIServer pid=145800) INFO:     Waiting for application startup.
(APIServer pid=145800) INFO:     Application startup complete.

Test Result

---

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.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.

[mergify]

Documentation preview: https://vllm--27974.org.readthedocs.build/en/27974/

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[youkaichao]

the direction looks good to me. thanks for doing it!

please take care of the correctness and make sure all tool parsers actually work.

you can also try to measure some import time after the refactor, and see if it helps.

[youkaichao]

the refactor is motivated by https://ppwwyyxx.com/blog/2023/Registration-Does-Not-Scale-Well/ . thanks to @ppwwyyxx