Manual2Skill++: Connector-Aware General Robotic Assembly from Instruction Manuals via Vision-Language Models