To add to Matt's answer, it is a generally accepted thing that numpy is often much slower on very small datasets and for simpler math problems. Both of which I think this test would qualify for. Travis Oliphant, the creator of numpy, seems to regularly address this and basically comes back to the fact that numpy will always beat out other solutions on much larger and more computationally intensive problems, but due to many factors like the ones Matt has highlighted will always be slower on very small and simple problems. Here is Travis addressing your question directly in an interview from a few years ago:
https://youtu.be/gFEE3w7F0ww?si=mfTO-uJQRIZdMKoL&t=6080