Add 5090 dry run stage to PR test workflow (#17022)

.github/workflows/pr-test.yml

@@ -672,6 +672,58 @@ jobs:
           fi
           python3 run_suite.py --hw cuda --suite stage-b-test-large-1-gpu $CONTINUE_ON_ERROR_FLAG
 
+  # 5090 dry run stage - only runs on scheduled CI or when explicitly targeted via /rerun-stage
+  # Used to validate 5090 compatibility before full integration
+  stage-b-test-small-1-gpu-5090:
+    needs: [check-changes, call-gate, stage-a-test-1, sgl-kernel-build-wheels]
+    if: |
+      always() &&
+      (
+        (inputs.target_stage == 'stage-b-test-small-1-gpu-5090') ||
+        (
+          !inputs.target_stage &&
+          github.event_name == 'schedule' &&
+          (!failure() && !cancelled()) &&
+          ((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
+        )
+      )
+    runs-on: 1-gpu-5090
+    continue-on-error: true
+    env:
+      RUNNER_LABELS: 1-gpu-5090
+      IS_BLACKWELL: "1"
+    strategy:
+      fail-fast: false
+      max-parallel: 8
+      matrix:
+        partition: [0, 1, 2, 3, 4, 5, 6, 7]
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ inputs.pr_head_sha || inputs.ref || github.sha }}
+
+      - name: Download artifacts
+        if: needs.check-changes.outputs.sgl_kernel == 'true'
+        uses: actions/download-artifact@v4
+        with:
+          path: sgl-kernel/dist/
+          merge-multiple: true
+          pattern: wheel-python3.10-cuda12.9
+
+      - name: Install dependencies
+        timeout-minutes: 10
+        run: |
+          source /etc/profile.d/sglang-ci.sh
+          CUSTOM_BUILD_SGL_KERNEL=${{needs.check-changes.outputs.sgl_kernel}} bash scripts/ci/ci_install_dependency.sh
+
+      - name: Run test
+        timeout-minutes: 30
+        run: |
+          source /etc/profile.d/sglang-ci.sh
+          cd test/
+          python3 run_suite.py --hw cuda --suite stage-b-test-small-1-gpu-5090 --auto-partition-id ${{ matrix.partition }} --auto-partition-size 8 --continue-on-error
+
   stage-b-test-large-2-gpu:
     needs: [check-changes, call-gate, stage-a-test-1, sgl-kernel-build-wheels]
     if: |