Numpy arrays of dtype=np.float128 get converted to Array{Nothing}

Float128 isn't one of the standard numeric types in Julia, but I don't think this behavior is intended.
```Julia
julia> py"np.array([1.,2.,3.,4.,5.], dtype=np.float128)"
5-element Array{Nothing,1}:
nothing
nothing
nothing
nothing
nothing
```
Oversized complex values are handled much more gracefully, by producing a PyObject.
```Julia
julia> py"np.array([1.,2.,3.,4.,5.], dtype=np.complex256)"
PyObject array([1.+0.j, 2.+0.j, 3.+0.j, 4.+0.j, 5.+0.j], dtype=complex256)
```