Just to confirm that I understood correctly: in this current bundled build, dfttest is functional and working on silicon Macs, but because it lacks native ARM optimizations, it's just not as fast as it theoretically could be?
Yes, the current dfttest, is compiled for arm neon, but afaik. has no specific optimizations (aside from those that the compiler does) and will be slower than one running on x86_64 systems utilizing AVX2&co.