Text this: Assessing DeepSeek R1 and ChatGPT 4.5 in Arabic-English literary translation: performance, challenges, and implications