Completeness of Unbounded Best-First Game Algorithms