Update kimik2.5-int4-mi300x-vllm vLLM ROCm image to v0.21.0 (#1404)

* Update kimik2.5-int4-mi300x-vllm vLLM ROCm image to v0.21.0

Ref #1154

The previous PR head was 59 commits behind main and reverted ~20
unrelated image bumps from other merged PRs. This branch was reset to
current main and the intended kimik2.5-int4-mi300x-vllm bump applied
fresh (v0.18.0 -> v0.21.0, matching the PR title; the prior commit
mis-targeted v0.20.2).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Re-trigger sweep (chi-mi300x-049 drained, healthy nodes only)

---------

Co-authored-by: claude-fix-bot <claude-fix-bot@local>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: functionstackx <47992694+functionstackx@users.noreply.github.com>

Author: 4 people

.github/configs/amd-master.yaml

@@ -546,7 +546,7 @@ kimik2.5-int4-mi325x-vllm:
       - { tp: 8, conc-start: 4, conc-end: 64 }
 
 kimik2.5-int4-mi300x-vllm:
-  image: vllm/vllm-openai-rocm:v0.18.0
+  image: vllm/vllm-openai-rocm:v0.21.0
   model: moonshotai/Kimi-K2.5
   model-prefix: kimik2.5
   runner: mi300x

perf-changelog.yaml

@@ -2623,3 +2623,9 @@
   description:
     - "Update vLLM image from v0.15.1 to v0.20.2"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1394
+
+- config-keys:
+    - kimik2.5-int4-mi300x-vllm
+  description:
+    - "Update vLLM ROCm image from v0.18.0 to v0.21.0"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1404