We had a test claiming that this empty struct type consumes a register
slot when passing it to a function with GCC, but that does not appear to
be the case, at least with GCC versions going back to 4.8.
This also caused a miscompilation when passing one of these structs to a
variadic function, but it turned out that our implementation of `va_arg`
matches GCC's ABI, so the one change fixes both bugs.
Similar to other targets (AMDGPU, Mips, PowerPC, RISCV, X86, ...)
`ninja check-clang-codegen-aarch64` can be used to test this subfolder.
Pull Request: https://github.com/llvm/llvm-project/pull/115818
