Implement write with internal buffer

Implement a buffered writer over an expensive sink API. You are given a function writeToDevice(byte[] chunk) that may accept at most M bytes per call and has high per-call overhead. Implement a class BufferedWriter with methods: write(byte[] data), flush(), and close(). Requirements: maintain byte order across multiple calls, minimize the number of writeToDevice calls, handle inputs larger than the internal buffer, and ensure no data loss on flush/close. Describe your buffer management, edge cases across consecutive writes, and the time/space complexity. Optionally discuss how you would add thread-safety.