fix(setup): improve precompiled wheel setup for Docker builds

[dougbtv]

This refactors the logic for handling VLLM_USE_PRECOMPILED to ensure that Docker builds extract only the required .so files and properly modify the package_data before setup()

• Removes errant precompiled wheel copy that was copying old code
  • e.g. not code from the current checkout.
• Moves precompiled wheel extraction logic into a utility class
• Applies package_data patch before calling setup()
• Now skips build_ext when precompiled is enabled
• Supports fallback to nightly wheel if latest commit isn't available

Follow-up to PR #21964 as part of improving build times for CI.

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

Fixes errant wheel copy and improves the flow so that package_data is computed prior to setup.

Test Plan

Build and inspect on local work station.

Test Result

Previously, old code would be overwritten by the copied wheel.

Locally built and inspected file contents.

Realized that the package_data list was compiled too late, refactored the flow to account for it, doing everything up front before setup()

Wound up fixing it up and inspecting final build like so to look for a modification I put in place:

doug@stimsonmt:~/vllm$ docker build --file docker/Dockerfile --build-arg max_jobs=16 --build-arg VLLM_USE_PRECOMPILED=1 --build-arg VLLM_DOCKER_BUILD_CONTEXT=1 --build-arg USE_FLASHINFER_PREBUILT_WHEEL=true --tag dougbtv/vllm:checkbuildfinal . | tee checkbuild.final.log.out
doug@stimsonmt:~/vllm$ docker run -it --rm --entrypoint=/bin/bash localhost/dougbtv/vllm:checkbuildfinal
root@35a10e831010:/vllm-workspace# cat /usr/local/lib/python3.12/dist-packages/vllm/config.py | grep -i bang
                "memory footprint and boosts the performance. !bang"

Found my marker in a modified file, e.g. !bang as seen in print output above.

(Optional) Documentation Update

[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 refactors the precompiled wheel handling for Docker builds, which is a good improvement for maintainability. The core logic is moved into a utility class, and the extraction process now happens before the main setup() call. I've identified a critical bug that would cause the setup script to crash and a high-severity resource leak from uncleaned temporary directories. Addressing these will make the build process more robust.

[mgoin]

@dougbtv @DarkLight1337 @simon-mo I think this broke the Python-only Installation Test https://buildkite.com/vllm/ci/builds/25598/steps/canvas?sid=01986167-695e-4a0e-b2a5-dd8aef0cb236