[CI] Reorganize release pipeline: separate nightly vs release sections

Reorder the pipeline into clear sections for readability:

1. Build Python Wheels (always runs)
2. ROCm Wheel Pipeline (always runs)
3. Build & Publish Docker Images (auto on nightly, blocked otherwise)
4. Release (manual) - version input, PyPI upload, CPU image builds,
   ROCm root index

Key changes:
- Extract CPU image builds (manual/blocked) from the image build group
  into their own "Build release CPU Docker images" group
- Move ROCm wheel jobs (1-4) up next to CUDA wheel builds
- Move release version input and PyPI publish to the release section

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.buildkite/release-pipeline.yaml

@@ -341,12 +341,18 @@ steps:
       S3_BUCKET: "vllm-wheels"
 
   # =============================================================================
-  # Nightly: Build & Publish Docker Images (NIGHTLY=1 only)
+  # Build & Publish Docker Images
+  # Automatically runs when NIGHTLY=1, otherwise blocked for manual trigger.
   # =============================================================================
 
-  - group: "Build nightly Docker images"
+  - block: "Unblock to build release Docker images"
+    key: block-build-release-images
+    if: build.env("NIGHTLY") != "1"
+
+  - group: "Build release Docker images"
     key: "build-release-images"
-    if: build.env("NIGHTLY") == "1"
+    depends_on: block-build-release-images
+    allow_dependency_failure: true
     steps:
       - label: "Build release image - x86_64 - CUDA 12.9"
         depends_on: ~
@@ -439,9 +445,8 @@ steps:
           - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=13.0.1 --build-arg UBUNTU_VERSION=24.04 --build-arg GDRCOPY_OS_VERSION=Ubuntu24_04 --build-arg FLASHINFER_AOT_COMPILE=true --build-arg torch_cuda_arch_list='8.7 8.9 9.0 10.0+PTX 12.0 12.1' --build-arg INSTALL_KV_CONNECTORS=true --build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.1-devel-ubuntu24.04 --tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m)-cu130-ubuntu2404 --target vllm-openai --progress plain -f docker/Dockerfile ."
           - "docker push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m)-cu130-ubuntu2404"
 
-  - group: "Publish nightly images"
+  - group: "Publish release images"
     key: "publish-release-images"
-    if: build.env("NIGHTLY") == "1"
     steps:
       - label: "Create multi-arch manifest - CUDA 12.9"
         depends_on:
@@ -503,6 +508,7 @@ steps:
       - label: "Publish nightly multi-arch image to DockerHub"
         depends_on:
           - create-multi-arch-manifest
+        if: build.env("NIGHTLY") == "1"
         agents:
           queue: small_cpu_queue_release
         commands:
@@ -520,6 +526,7 @@ steps:
       - label: "Publish nightly multi-arch image to DockerHub - CUDA 13.0"
         depends_on:
           - create-multi-arch-manifest-cuda-13-0
+        if: build.env("NIGHTLY") == "1"
         agents:
           queue: small_cpu_queue_release
         commands:
@@ -534,11 +541,12 @@ steps:
           DOCKER_BUILDKIT: "1"
           DOCKERHUB_USERNAME: "vllmbot"
 
-  # ROCm nightly Docker image
+  # ROCm Docker image (also gated behind the block step)
   - label: ":docker: Build release image - x86_64 - ROCm"
     id: build-rocm-release-image
-    if: build.env("NIGHTLY") == "1"
     depends_on:
+      - step: block-build-release-images
+        allow_failure: true
       - step: build-rocm-base-wheels
         allow_failure: false
     agents:
@@ -685,6 +693,9 @@ steps:
         env:
           DOCKER_BUILDKIT: "1"
 
+  # ROCm Job 5: Generate Root Index for ROCm Wheels (for release only)
+  # This is the job to create https://wheels.vllm.ai/rocm/ index allowing
+  # users to install with `uv pip install vllm --extra-index-url https://wheels.vllm.ai/rocm/`
   - block: "Generate Root Index for ROCm Wheels for Release"
     key: block-generate-root-index-rocm-wheels
     depends_on: upload-rocm-wheels