Support for THn and THnSparse in hist2array

[ndawe]

As requested by @cbourjau in #188 this PR adds support for THn and THnSparse (with #arbitrary dimensionality) in hist2array:

>>> from root_numpy import array2hist, hist2array
>>> import numpy as np
>>> from rootpy.plotting import Hist2D
>>> hist = Hist2D(5, 0, 1, 3, 0, 1, type='F')
>>> array = np.random.randint(0, 10, size=(7, 5))
>>> array
array([[0, 4, 3, 5, 2],
       [2, 4, 4, 1, 2],
       [8, 1, 3, 4, 8],
       [1, 3, 0, 2, 3],
       [1, 8, 5, 6, 8],
       [4, 7, 8, 4, 5],
       [8, 7, 1, 4, 5]])
>>> array2hist(array, hist)
Hist2D('Hist2D_EJWHJJNmSmHPVeeiWtzFkT')
>>> hist2array(hist)
array([[ 4.,  4.,  1.],
       [ 1.,  3.,  4.],
       [ 3.,  0.,  2.],
       [ 8.,  5.,  6.],
       [ 7.,  8.,  4.]], dtype=float32)
>>> import ROOT
>>> hist2array(ROOT.THn.CreateHn("", "", hist))   # <=== same result
array([[ 4.,  4.,  1.],
       [ 1.,  3.,  4.],
       [ 3.,  0.,  2.],
       [ 8.,  5.,  6.],
       [ 7.,  8.,  4.]], dtype=float32)
>>> hist2array(ROOT.THnSparse.CreateSparse("", "", hist))   # <=== same result
array([[ 4.,  4.,  1.],
       [ 1.,  3.,  4.],
       [ 3.,  0.,  2.],
       [ 8.,  5.,  6.],
       [ 7.,  8.,  4.]], dtype=float32)