[llvm] [CodeGen][AArch64] Set min jump table entries to 13 for AArch64 targets (PR #71166)

[David Sherwood via llvm-commits]

david-arm wrote:

> How does this affect performance in cases where the last switch case is usually taken?

Honestly speaking I don't know! As mentioned in the commit message, I just changed this number until I got the best performance results, so it's quite possible that in some benchmarks that I've not tested this does make things worse. If we discover issues later on we can revisit this and perhaps refine the algorithm.

https://github.com/llvm/llvm-project/pull/71166