docs: publish BU Bench benchmark report

[beubax]

Description

Publishes the 100-task BU Bench results in the main README and adds a detailed, research-style benchmark report.

• Adds a compact comparison table to the main README, placed above Learn More
• Adds separate SVG charts for accuracy, total tokens, API-equivalent cost, and agent turns
• Explains the code-based browser executor, snapshot pruning, and task-aware diff policy
• Documents category results, the evaluation setup, judge rubric, limitations, and reproduction steps
• Replaces the previous combined benchmark chart

Related issue: N/A

Type of Change

• 🐛 Bug fix
• ✨ New feature
• 🌐 New site adapter
• 📝 Documentation
• ♻️ Refactor
• 🔧 CI / build / tooling

Checklist

• I ran the checks relevant to this PR
• I updated tests or docs if needed
• I included output or screenshots when useful
• If I edited skill-src/, I ran make build and committed skills/

Checks

• git diff --check origin/main...HEAD
• xmllint --noout benchmarks/charts/*.svg
• README metric, placement, and local-link assertions
• uv run --project benchmarks python benchmarks/scripts/run_eval.py --help

The application test suite was also attempted after rebasing. All assertions passed except one unrelated upstream programmatic-runner isolation test that timed out locally; this PR changes documentation and SVG assets only.

Screenshots / Output

The benchmark report embeds the four SVG charts directly beneath their corresponding metrics.

[github-actions]

🟢 No documentation gap found — medium confidence

The automated review found no documentation gap in the supplied changes.

This review is advisory and does not block merging.