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.