Add REFUTE scientific critique + calibration benchmark

[connerlambden]

Proposal

Add REFUTE to related / evaluation / research tooling docs if this project surfaces LLM evaluation, RAG quality, or scientific agent workflows.

REFUTE is a scientific critique + calibration benchmark (paper-grounded claims → predictions → judge scores → Brier/ECE).

• Product: https://bgpt.pro/refute
• Inspect task: https://github.com/connerlambden/refute-inspect
• Interim preprint release: https://github.com/connerlambden/refute-inspect/releases/tag/v3.0.0-preprint

Happy to adjust wording / PR if preferred.