feat: Add step-length metrics for measuring Claude's autonomy

[mylee04]

Summary

I saw the discussion in the comments about tracking step-length metrics, so I decided to give it a try! This PR implements the ability to track consecutive tool-using commands before interruption, which helps understand how autonomously Claude operates.

What it does

• Tracks how many consecutive commands use tools before being interrupted
• Shows distribution of step lengths (e.g., how often Claude takes 1, 2, 3+ steps)
• Displays trends over time to see if Claude is becoming more autonomous
• Calculates average, min, and max step lengths

Implementation

• Backend: Added step-length calculation in stats.py
• Frontend: Added two new charts in the dashboard
  • Step-Length Distribution (bar chart)
  • Step-Length Metrics Over Time (line chart)
• Tests: Added comprehensive test suite (8 tests, all passing)

Example metrics

{
  "average_step_length": 2.38,
  "max_step_length": 4,
  "step_length_distribution": {"1": 2, "2": 3, "3": 1, "4": 2}
}

The charts appear after the Interruption Rate chart in the dashboard and automatically hide when there's no data.

Let me know if you'd like any changes or have feedback!

[chiphuyen]

Thanks for the contribution!

Can you share a screenshot of what it might look like?

I did consider adding a similar chart, but then I realized it makes the chart page a bit crowded. I think a good feature would be to let users customize what set of charts they want for each project.

Currently, I think it might also look similar to the User Commands chart, which maps the distributions of steps per command.

[mylee04]

@chiphuyen You are very welcome! Yeah you're correct. It looks similar to user command.

[chiphuyen]

Thanks for the follow-up. I'll wait to merge until we have custom chart feature (to avoid making the dashboard crowded), but I'll keep the PR up here in case folks want to use this chart!