std() crashes for multidimensional arrays when no axis is specified

[terhorst]

>> A = np.array([[1,2,3],[4,5,6]]); A
array([[1, 2, 3],
       [4, 5, 6]])

>> A.std()
1.707825127659933

>> B = DataArray(A, 'xy'); B
DataArray([[1, 2, 3],
       [4, 5, 6]])
('x', 'y')

>> B.std('x')
DataArray([ 1.5,  1.5,  1.5])
('y',)

>> B.std()
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
/Users/terhorst/Documents/Work/datarray/build/lib/ in ()
----> 1 B.std()

/Users/terhorst/Documents/Work/datarray/build/lib/datarray/datarray.py in runs_op(*args, **kwargs)
    518             # do nothing special if not a DataArray, otherwise

    519             # this is a full reduction, so we lose all axes

--> 520             return super_op(inst, **kwargs)
    521 
    522         axes = list(inst.axes)

/Users/terhorst/Documents/Work/datarray/build/lib/datarray/datarray.py in __array_prepare__(self, obj, context)
    688                 this_ax = these_axes.pop(-1)
    689                 # print self.shape # dbg

--> 690                 this_dim = self.shape[this_ax.index]
    691                 that_dim = other.shape[that_ax.index]
    692                 if that_ax.name != this_ax.name:

IndexError: tuple index out of range

[fperez]

Hey, thanks for jumping on this so quick!