[lld] [lld] Properly diagnose R_AARCH64_PREL32 overflow (PR #205963)

[Peter Smith via llvm-commits]

smithp35 wrote:

I've asked internally whether anyone has any memory of why R_<CLS>_PCREL32 is defined that way. If there's no good reason, given BFD's implementation I think there's a good case to update the ABI. Will hopefully have an answer this week.

https://github.com/llvm/llvm-project/pull/205963