[CI] upload AIPerf Perfetto traces from benchmark workflows

[MengqingCao]

Summary

• Convert AIPerf profile_export.jsonl into gzipped Chrome traces after benchmark runs, and upload them as workflow artifacts with Job Summary download links for Perfetto.
• Share the converter through .github/actions/upload-aiperf-traces and wire it into ATOMesh, native ATOM, vLLM, and SGLang benchmark workflows.
• Conversion is best-effort: missing JSONL is a no-op, conversion errors warn and do not fail the benchmark.

Motivation

Agentic AIPerf runs already write per-request profile_export.jsonl, but inspecting concurrency slots / TTFT / wait∩prefill currently requires a local conversion step. CI should produce a Perfetto-ready trace and a download link on the run page.

Native / vLLM / SGLang benches today use bench_serving summary JSON, so they will skip this step until they emit profile_export.jsonl. The shared action is already in those workflows so no extra wiring is needed later.

Test plan

• Mark the PR ready (or workflow_dispatch Atomesh Benchmark) and run one GLM agentic case, e.g. glm-52-mxfp4-1p1d-tp4-agentic-1m-c1
• Confirm the job uploads aiperf-chrome-traces-* and the Job Summary has a Perfetto download link
• Download .trace.json.gz, open in https://ui.perfetto.dev (Open trace file), and check slot / TTFT / prefill+decode tracks

Test result

test passed in https://github.com/ROCm/ATOM/actions/runs/34118300963/job/101730333679, the trace could be uploaded and downloaded normally.

Submission Checklist

• Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

[github-actions]

🏷️ CI Guide

Runs automatically on every eligible PR before approval:

• ✅ Pre Checkin: Black, Ruff, catalog schema validation, non-GPU unit tests

Heavy model tests:

• ✅ Run after the PR is approved and Pre Checkin passes
• ✅ Run immediately when an approval review is submitted
• ✅ Can be requested before approval with labels

Label	Tests
ci:full	Run all heavy PR model tests: native ATOM, vLLM, and SGLang
ci:atom	Run native ATOM model accuracy tests
ci:vllm	Run ATOM vLLM OOT model accuracy tests
ci:sglang	Run ATOM SGLang model accuracy tests

Heavy jobs are skipped when the PR is not approved and no matching ci:* label is present.
Add labels via the sidebar or gh pr edit 2155 --add-label <label>

[Jasen2201]

LGTM