openai/gpt-chat-latest → A (4:1)

Side A fixes a fundamental correctness bug in the ranking algorithm by switching Rank Centrality to the canonical degree-based d_max, eliminating oscillation in star-topology graphs and producing the correct stationary distribution. It also adds focused regression tests (Rust and end-to-end fixture tests) that lock in the behavior, whereas Side B mixes a real external index fix with a large refactor, GitHub card rendering, module moves, and UI enhancements whose lasting value is broader but less foundational.