~x-ai/grok-latest → A (2:1)

A fixes a real correctness bug in the core Rank Centrality math (wrong d_max caused bipartite oscillation and uniform star scores), aligns the implementation with the paper, and adds targeted regression tests. B is a worthwhile structural simplification (votes-only ScopeVotes, derive edges on demand) but is largely mechanical renames and state-hoisting across many call sites without changing ranking behavior.