[llvm] [llvm] Optimize print-on-crash serialization (PR #205682)

[Alexis Engelke via llvm-commits]

aengelke wrote:

Is this still faster with #216838 (and possibly further optimizations for printing)? Can you provide more detailed performance numbers that indicate why printing is slow? I'm still in the camp of trying to make printing faster (esp. no longer O(n) in module size per function) before trying alternative measures.

https://github.com/llvm/llvm-project/pull/205682