Revert "feat: add perf comparison on non-main brances with run-sweep workflow" [skip-sweep]

[cquil11]

Reverts #880

[github-actions]

Thanks for the contribution! For vLLM & SGLang, please ensure that your recipes is similar to the official vLLM recipes and/or the SGLang cookbook

• https://cookbook.sglang.io/
• https://docs.vllm.ai/projects/recipes/en/latest/

If it is not, please create a PR first before we can merge your PR into the master branch. Let's ensure that the documentation is first class such that the entire ML community can benefit from your hard work! Thank you

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sgl-cookbook

[claude]

LGTM — clean revert of #880 with no bugs found.

Extended reasoning...

Overview

This PR reverts #880, which added a compare-results CI job and utils/compare_results.py script for benchmarking performance comparisons on non-main branches. The revert touches four files: the GitHub Actions workflow (run-sweep.yml), the comparison script (utils/compare_results.py), a config file (nvidia-master.yaml), and perf-changelog.yaml.

Security risks

None. The removed code queried a read-only database URL via an environment secret. Removing it reduces the attack surface rather than expanding it. No auth, crypto, or permission-sensitive code is introduced.

Level of scrutiny

Low — this is a mechanical revert. The workflow job removal is a straightforward deletion, the script deletion removes previously added code, the config restoration uncomments previously commented-out YAML search-space entries, and the changelog cleanup removes a placeholder "test" entry. All changes are consistent with reverting #880.

Other factors

No bugs were found by the automated bug hunting system. The PR is small, focused, and carries the [skip-sweep] tag to prevent unnecessary benchmark runs. The only timeline activity is a boilerplate bot comment unrelated to the code.