Spark 4.1: Optimize UUID writer with fast-path byte parsing.

[slfan1989]

Summary

• Optimize UUID writer to avoid UTF-8 decoding and string parsing overhead
• Add fast-path for canonical UUID format (8-4-4-4-12 with hyphens)
• Direct byte-level parsing from UTF8String to ByteBuffer

[slfan1989]

@nastra @huaxingao Could you please review this PR when you have a moment? Thank you very much!

[singhpk234]

@slfan1989 do you perf benchmarks on how much it improves ? I wonder if its worth the complexity

[slfan1989]

@slfan1989 do you perf benchmarks on how much it improves ? I wonder if its worth the complexity

@singhpk234 Thanks for your reply. I’ll try to provide a test report. In theory, there should be a performance improvement.