perf(amd): port MI355X Qwen3.5 MXFP4 disagg sweep to srt-slurm

[cquil11]

Summary

• replace the active MI355X Qwen3.5 MXFP4 legacy AMD disaggregated launcher with native srt-slurm orchestration
• preserve the production 1P1D TP8 topology and complete c8/c16/c32/c64/c128/c256/c512 8k1k search space
• use the current steady SGLang ROCm v0.5.17 image, SGLang Router, AMD MoRI, AITer attention, and FP8 KV cache
• call the repository's unchanged benchmark_serving.py directly through the custom benchmark contract
• remove the model-specific wrapper that only translated matrix fields into amd_utils/submit.sh

Validation

• srt-slurm dry-run passes against feat(runtime): add AMD accelerator support srt-slurm#1 at c609754b5622f96d5c12a93149e245308d4f1e9b
• Hugging Face model identity resolves and the pinned 64 GiB SGLang squashfs is already present on MI355X
• generated matrix contains exactly one topology job; the recipe owns all seven concurrency points
• changed YAML parses and git diff --check passes

Stack

Based directly on #2542. The shared launcher and cluster profile live only in the base; this PR contains only the Qwen3.5 MXFP4 matrix entry, recipe, and changelog entry. This remains draft until the exact-head MI355X recipe sweep passes.

[github-actions]

Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase As a PR reviewer and CODEOWNER, I have reviewed this and have.

For PR verification, add the full-sweep-fail-fast label (strongly recommended) to this PR — the benchmark sweep only runs on labeled PRs. Use full-sweep-enabled only if you need matrix jobs to keep running past a failure.

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. See GitHub's docs on re-running failed jobs

---

感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 As a PR reviewer and CODEOWNER, I have reviewed this and have。

如需进行 PR 验证，请为此 PR 添加 full-sweep-fail-fast 标签（强烈推荐）— 基准测试 sweep 仅在带有标签的 PR 上运行。仅当需要矩阵任务在失败后继续运行时才使用 full-sweep-enabled。

PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。参见 GitHub 关于重新运行失败任务的文档