Have floating point functions take their parameters by value.

Make all of the methods in `std::num::Float` take `self` and their other parameters by value.

Some of the `Float` methods took their parameters by value, and others took them by reference. This standardises them to one convention. The `Float` trait is intended for the built in IEEE 754 numbers only so we don't have to worry about the trait serving types of larger sizes.

[breaking-change]

src/librand/distributions/gamma.rs

@@ -147,7 +147,7 @@ impl IndependentSample<f64> for GammaSmallShape {
     fn ind_sample<R: Rng>(&self, rng: &mut R) -> f64 {
         let Open01(u) = rng.gen::<Open01<f64>>();
 
-        self.large_shape.ind_sample(rng) * u.powf(&self.inv_shape)
+        self.large_shape.ind_sample(rng) * u.powf(self.inv_shape)
     }
 }
 impl IndependentSample<f64> for GammaLargeShape {