[Klaud Cold] refactor: remove SWE-bench Lite eval suite

[functionstackx]

Summary

Remove the SWE-bench Lite eval suite and every reference to it across the codebase.
SWE-bench Lite is unused -- no master config selects it, and the eval framework has moved past it.

Deleted files (5)

• infx/evals/swebench_lite.yaml -- task definition
• infx/evals/swebench_score.py -- scorer
• infx/evals/patches/patch_swebench_agent.py -- agent runtime patch
• infx/evals/patches/patch_swebench_scoring.py -- scoring runtime patch
• infx/tests/evals/test_swebench_eval.py -- dedicated test module

Edited files (28)

• Workflows: benchmark-tmpl.yml, benchmark-multinode-tmpl.yml, e2e-tests.yml -- removed swebench inputs, env vars, secrets, artifact globs
• Shell scripts: benchmark_lib.sh (removed ~280-line swebench function + dispatch case), runtime_settings.sh, srt_eval.sh
• Launchers: launch_h100-cr.sh, launch_mi325x-tw.sh -- removed SWEBENCH_/MODAL_ docker env passthrough
• Slurm infra: slurm_utils.sh, amd_utils/job.slurm, amd_utils/submit.sh, llm-d/job.slurm, llm-d/submit.sh
• Tests: test_run_eval_dispatch.py (~470 lines of swebench/modal tests removed + unused signal import), test_eval_patches.py (dead _load_patch_module helper + unused imports), test_generate_sweep_configs.py (docstring)
• Docs (EN+ZH): architecture.md, ci-procedures.md, eval-agentx-procedures.md, index.md, results-and-ingestion.md and their _zh.md counterparts
• Eval docs: EVALS.md -- removed SWE-bench Lite section, patch entries, task file reference
• Thresholds: thresholds.yaml -- removed swebench_lite: 0.50 entry

Deps removed

• Modal credential plumbing (MODAL_TOKEN_ID / MODAL_TOKEN_SECRET) was exclusively consumed by swebench and is removed everywhere.
• No pyproject.toml / lock changes needed (mini-swe-agent and swe-rex were never declared as deps).

Validation

• bash -n passes on all touched .sh / .slurm files
• uvx ruff check infx/ -- all checks passed
• pytest infx/tests/evals/ infx/tests/matrix/ runners/ -- 607 passed
• YAML workflow parsing -- all 4 files valid
• git grep -n -i -E 'swe.?bench|mini.?swe|SWEBENCH' -- . ':!perf-changelog.yaml' -- zero hits

Test plan

• CI e2e-tests workflow passes (no swebench env vars expected)
• Benchmark template workflows validate (no swebench inputs/secrets referenced)
• Existing non-swebench evals (lm-eval, BFCL, AgentX, vendor) unaffected
• Multi-node slurm jobs launch without referencing removed env vars

Follow-up (41ea1b9)

• .github/workflows/run-sweep.yml: stop passing MODAL_TOKEN_ID / MODAL_TOKEN_SECRET to benchmark-tmpl.yml / benchmark-multinode-tmpl.yml (12 call sites, 24 lines). The templates no longer declare those secrets, and GitHub rejects reusable-workflow calls that pass undeclared secrets, so every sweep would have failed to start. No MODAL references remain under .github/. The MODAL_TOKEN_* repo secrets themselves can be deleted separately.

🤖 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。

[adibarra]

Heads-up: #3576 (merged) replaced the bash launchers with a Python launcher, so this PR will conflict when you merge main, and the sweep won't start until that's resolved. Please merge main and move your launcher changes over to configs/runners.yaml / infx/launch/. Apologies for the churn, and thanks for your understanding as we wrap up the repo-wide refactoring push.