[Cryptography] Proper Entropy Source

[Ryan Carboni]

>Specifically:  A RNG that fails a statistical test is
>surely broken, but one that passes such a test is not
>guaranteed to be any good.  h/t Dykstra.

To circle back to my original post, the FFT was used to detect
patterns by a Cloudflare employee in RAM refresh.

If there are cyclical patterns, a FFT would detect some of them.

A RNG that fails a statistical test may not be broken if the problem
can be worked around. Otherwise you're all stupid and I should stop
using Linux, and maybe go back to using rock and sticks.