Don't run doctests for FloatCore::{min, max} methods This is to mitigate Debian bug #950337. The doctests show how to use the `min` and `max` methods of the `FloatCore` trait, but with the `std` feature enabled, they fall back to the implementations built into the compiler, which shows different behavior on `mipsel` and `mips64el` architectures than on the other architectures supported by Debian. The bug in the compiler is tracked in Debian bug #950583. As soon as that is fixed, this patch should be removed.
