Add method for multiply-add aka axpy

[bluss]

Two reasons:

• Blas offers this as axpy, so it can be optimized (note: axpy is for vectors, not strided matrix)
• It's a common case.

This operation is already _available efficiently_ using .zip_mut_with:

a1.zip_mut_with(&a2, |x, &y| *x += k * y)

The question is if this is so common and the zip_mut_with so ugly that it warrants its own method.

[bluss]

This interface is actually cool:

        self.w += Scaled(-step, &dt1);
        self.y += Scaled(-step, &dt2);

The problem is that if we provide AddAssign for Scaled, we'd probably better implement all operators for Scaled, and that quickly grows out of proportion..