nrf seems not to use the micropython math library

[jepler]

I noticed this, and @dhalbert said

that is a mistake, we should probably use the micropython one to consistent results (even if they are less accurate or whatever)

[jepler]

I was mistaken. CircuitPython centralizes the choice, so I misinterpreted what was going on in some merge conflicts in #4897.