Sampling from von Mises Fisher yields NaN

This happens when the mean direction unit vector `μ` is along the x-axis. For instance:
```
julia> vMF = VonMisesFisher([1, 0], 1.0)
VonMisesFisher{Float64}(μ=[1.0, 0.0], κ=1.0)
julia> rand(vMF)
2-element Vector{Float64}:
 NaN
 NaN
```
This problem seems to persist in higher dimensions.