Add SWTBench profiling timelines

[simonrosenberg]

Summary

• add per-instance attempt timelines in evaluation orchestrator
• capture SWTBench eval/harness timings and emit structured JSON timelines
• inject profiling sitecustomize into swt-bench at runtime (optional via PROFILE_SWTBENCH)

Testing

• not run (profiling-only instrumentation)

Note: vendor/software-agent-sdk submodule advanced to current main when updating base branch; no manual edits inside submodule.

[openhands-ai]

Looks like there are a few issues preventing this PR from being merged!

• GitHub Actions are failing:
  • Pre-commit checks

If you'd like me to help, just leave a comment, like

@OpenHands please fix the failing actions on PR #313 at branch `feature/swtbench-profiling`

Feel free to include any additional details that might help me get this PR into a better state.

_{^{You can manage your notification settings}}

[simonrosenberg]

@juanmichelini perhaps this could NOT be merged (because a bit messy) but used as a hack when we need to profile and optimize so branch... We can just ask OH to merge this into any other feature branch and then we get the profiling logs for free.. But I do enjoy the idea of adding precise profiling logs into the report.jsons.
What is less good is that the profiling code depends on each benchmark and Sitecustomize is hacky and messy so idk

[juanmichelini]

@juanmichelini perhaps this could NOT be merged (because a bit messy) but used as a hack when we need to profile and optimize so branch... We can just ask OH to merge this into any other feature branch and then we get the profiling logs for free.. But I do enjoy the idea of adding precise profiling logs into the report.jsons. What is less good is that the profiling code depends on each benchmark and Sitecustomize is hacky and messy so idk

Tough choice! Thoughs on doing a PR to the SWT bench repo?