Bug when building a view of a masked array with field offsets

[eduble]

When building a view of a masked structured array, using a dtype with offsets, a ValueError exception is fired.

Note: With an array of type np.array instead of np.ma.MaskedArray, all runs fine.

Reproducing code example:

import numpy as np
x = np.array([('Rex', 9, 81.0), ('Fido', 3, 27.0)], dtype=[('name', 'U10'), ('age', 'i4'), ('weight', 'f4')])
age_dtype, age_offset = x.dtype.fields['age']
newdt = np.dtype({ 'names': ['age'], 'formats': [age_dtype], 'offsets': [age_offset], 'itemsize': x.dtype.itemsize })
# this works for now; it gives a view of x with column 'age' only
x.view(newdt)
# now converting x to a masked array
x = x.view(np.ma.MaskedArray)
# Oops!
x.view(newdt)

Error message:

Traceback (most recent call last):
  File "/home/etienne/sakura/.venv/lib/python3.5/site-packages/numpy/ma/core.py", line 3130, in view
    if issubclass(dtype, ndarray):
TypeError: issubclass() arg 1 must be a class

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/etienne/sakura/.venv/lib/python3.5/site-packages/numpy/ma/core.py", line 3136, in view
    output = ndarray.view(self, dtype)
  File "/home/etienne/sakura/.venv/lib/python3.5/site-packages/numpy/ma/core.py", line 3357, in dtype
    self._mask.shape = self.shape
ValueError: cannot reshape array of size 6 into shape (2,)

Numpy/Python version information:

>>> print(sys.version)
3.5.2 (default, Nov 23 2017, 16:37:01) 
[GCC 5.4.0 20160609]
>>> print(np.__version__)
1.15.0
>>>