feat: enhance CI pipeline with improved caching and multi-arch support

[Aias00]

What type of PR is this?

enhance CI pipeline with improved caching and multi-arch support

What this PR does / why we need it:

Which issue(s) this PR fixes:

Fixes #359

Release Notes: Yes/No

[netlify]

✅ Deploy Preview for vllm-semantic-router ready!

Name	Link
🔨 Latest commit	0a509b7
🔍 Latest deploy log	https://app.netlify.com/projects/vllm-semantic-router/deploys/68e795dda55124000875981d
😎 Deploy Preview	https://deploy-preview-360--vllm-semantic-router.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[github-actions]

👥 vLLM Semantic Team Notification

The following members have been identified for the changed files in this PR and have been automatically assigned:

📁 Root Directory

Owners: @rootfs, @Xunzhuo
Files changed:

• Dockerfile.extproc.cross
• .github/workflows/docker-publish.yml
• .github/workflows/k8s-integration-test.yml
• .gitignore
• Dockerfile.extproc

📁 candle-binding

Owners: @rootfs
Files changed:

• candle-binding/Cargo.lock

---

🎉 Thanks for your contributions!

This comment was automatically generated based on the OWNER files in the repository.

[rootfs]

@yuluo-yx @Xunzhuo PTAL, thanks

[Copilot]

Pull Request Overview

This PR enhances the CI pipeline by improving Docker build caching strategies and adding multi-architecture build support. The changes optimize build times through better layer caching and provide flexible multi-arch builds that can be skipped for PR builds to speed up CI feedback.

• Adds GitHub Actions cache integration for Docker builds with cache-from and cache-to directives
• Implements conditional multi-architecture support that skips arm64 builds for PRs while maintaining full multi-arch for releases
• Optimizes Dockerfile layers by separating dependency downloads from source code copying

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
Dockerfile.extproc	Optimizes build layers with separate Go dependency download step and adds build optimizations
.github/workflows/docker-publish.yml	Adds GitHub Actions caching, conditional multi-arch builds, and PR trigger paths

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

The comment is misleading. The logic actually skips multi-arch for PRs unless explicitly enabled (when skip_multiarch is false), not 'unless explicitly disabled'.

Suggested change

	# Skip multi-arch for PRs to speed up CI, unless explicitly disabled
	# Skip multi-arch for PRs to speed up CI, unless explicitly enabled

[Copilot]

Hard-coding CARGO_BUILD_JOBS to 4 may not be optimal for all runner types. Consider using a dynamic value based on available CPU cores, such as CARGO_BUILD_JOBS=${{ github.event_name == 'pull_request' && '2' || '4' }}.

Suggested change

	CARGO_BUILD_JOBS=4
	CARGO_BUILD_JOBS=${{ github.event_name == 'pull_request' && '2' || '4' }}

[Aias00]

@yuluo-yx @Xunzhuo PTAL, thanks

hi, this pr is still in working

[Aias00]

@rootfs Hi, I believe I have completed this task. The build duration of the ARM architecture Docker image has been reduced to 20 minutes. I attempted to push a Docker image as part of the pull request workflow, but the PR does not have the necessary permissions to push images. If you have a moment, could you please review it again?

[rootfs]

@Aias00 thanks! merging it now for image build

[yuluo-yx]

hi @Aias00 Why submit this document?

[Aias00]

Cargo.lock enables Docker layer caching optimization by ensuring deterministic dependency resolution. When Cargo.lock remains unchanged between builds, Docker can reuse the cached dependency compilation layer, significantly reducing build times from ~6.5 minutes to ~1.2 minutes for source code changes.