[Klaud Cold] Delete benchmarks/multi_node/deprecated

[functionstackx]

Summary

Delete benchmarks/multi_node/deprecated/ (12 retired multi-node scripts: DSv4 GB200 llm-d / MI355X ATOM+SGLang disagg, GLM-5 MI355X SGLang, gpt-oss GB200 Dynamo-TRT, Kimi-K2.5 MI355X ATOM+vLLM, MiniMax-M2.5/M3 MI355X ATOM+vLLM disagg). Continues the cleanup from #3460/#3461/#3463/#3464: retired assets are deleted, with git history as the record.

Why this is safe

• No launcher, workflow, infx/, utils/ or recipe references multi_node/deprecated, and no active master config resolves any of these filenames (multi-node launchers look in benchmarks/multi_node/, not deprecated/).
• Two text references are fixed:
  • docs/configuration-procedures(_zh).md "Sources" line linked an llmd-vllm benchmark wrapper at benchmarks/multi_node/dsv4_fp4_gb200_llmd-vllm-disagg.sh, a path that already did not exist (the script lived only in deprecated/); the link is dropped.
  • benchmarks/multi_node/agentic/dsv4_fp4_mi355x_atom-disagg.sh header comment now calls its SGLang sibling "former" (comment only).

Test plan

• git grep finds no reference to multi_node/deprecated; bash -n on the edited agentic script.
• No config or perf-changelog.yaml change, so no sweep is triggered.

🤖 Generated with Claude Code

[github-actions]

Thanks for the contribution!

• Review: If this PR changes files owned by someone other than a repository admin or @SemiAnalysisAI/core, ask one eligible CODEOWNER to complete the latest PR_REVIEW_CHECKLIST.md before contacting a core maintainer on Slack. Follow the template exactly, including As a PR reviewer and CODEOWNER, I have reviewed this and have, so sign-off verification triggers.
• PR verification: Sweeps only run on labeled PRs. Add full-sweep-fail-fast (strongly recommended); use full-sweep-enabled only when matrix jobs should continue after a failure.
• After merging: PR authors must ensure all GitHub Actions jobs pass. Transient failures often pass on rerun; see how to rerun failed jobs.

中文

感谢你的贡献！

• **审阅：**如果 PR 修改的文件归属于仓库管理员及 @SemiAnalysisAI/core 之外的 CODEOWNER，请先联系一位有资格的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，再通过 Slack 联系核心维护者。必须严格遵循模板，并保留 As a PR reviewer and CODEOWNER, I have reviewed this and have，才能触发签核验证。
• **PR 验证：**扫描仅在带有标签的 PR 上运行。强烈建议添加 full-sweep-fail-fast；仅当需要矩阵任务在失败后继续运行时才使用 full-sweep-enabled。
• **合并后：**PR 作者必须确保所有 GitHub Actions 任务通过。临时性失败通常可以通过重新运行恢复；参见重新运行失败任务的说明。

[claude]

Looks good, straightforward deletion of deprecated benchmark scripts. Reviewed the file deletions (12 retired multi-node scripts, no other content changes), confirmed via search that no remaining references to multi_node/deprecated exist in the repo, verified the two doc link fixes and the one comment tweak match the stated intent, and checked that the deleted script filenames only appear elsewhere in a historical perf-changelog.yaml log entry (not an active reference).

Extended reasoning...

Change is a pure deletion of 12 unused deprecated benchmark shell scripts plus two documentation link fixes and one comment update in a sibling script; no code logic, auth, or security-sensitive paths are touched. Confirmed via repo-wide search that no remaining code references the deleted paths, matching the PR description's safety claim. No CODEOWNERS concerns apparent and the diff is small and mechanical, so a human need not review this.