Add rolling benchmark history tracking

[benvinegar]

Summary

• add a benchmark:history pipeline that resolves each benchmark repo's default-branch commit as of a recorded timestamp, scans it, and upserts one JSONL datapoint per repo per UTC week
• generate rolling history artifacts under benchmarks/history/known-ai-vs-solid-oss/ plus a markdown summary report at reports/known-ai-vs-solid-oss-history.md
• add mini sparkline trends to the rolling history report and keep compact series data in latest.json for downstream consumers
• show both the latest pinned score and the highest pinned score in the history tables so repos can be compared against their own prior peaks
• mirror that trend view in the README benchmark table with latest/highest/delta columns and links to the pinned and rolling benchmark artifacts
• support honest backfills with --recorded-at, skipping repos that did not exist yet instead of fabricating datapoints
• add a weekly GitHub Actions workflow to refresh benchmark history automatically and commit updates back to the repo
• seed the current benchmark set with a 4-week backfill of history datapoints
• tighten the top-level README by removing redundant sections while keeping supported file extensions explicit

Validation

• bun run format:check
• bun run lint
• bun test
• bun run build
• bun run benchmark:history --recorded-at 2026-03-23T12:00:00Z
• bun run benchmark:history --recorded-at 2026-03-30T12:00:00Z
• bun run benchmark:history --recorded-at 2026-04-06T12:00:00Z
• bun run benchmark:history --recorded-at 2026-04-13T12:00:00Z

This PR description was generated by Pi using GPT-5