Full-Sentence Models Perform Better in Simultaneous Translation Using the Information Enhanced Decoding Strategy