Mind the Gap: A Closer Look at Tokenization for Multiple-Choice Question Answering with LLMs – Takara TLDR