https://github.com/llvm/llvm-project/pull/74764 reported that the `lljit-with-remote-debugging` test fails on AArch64 hosts, because the input IR file states arch x86_64 explicitly. In order to drop the target triple we have to remove a check in the example implementation. Not sure it's fully portable now, but at least it's better than before.