openai/gpt-chat-latest → B (10:1)

Side B performs a substantial type-safety refactor by changing `resolve_item` to return `CanonicalItemUrl`, propagating canonical URL newtypes throughout validation, ranking, RPC, and connectivity code, and adding `Deref<Target=str>` implementations plus tests. This removes repeated string wrapping/parsing, reduces opportunities for mixing raw strings with canonical IDs, and improves API correctness, whereas Side A only rewrites conditional button rendering to emit a literal `disabled` attribute instead of a bound boolean.