Indexing an array with an object dtype axis using an Array fails

[gdementen]

Very high priority, because it is a regression and prevents Pension from working on latest larray version.

>>> a = Axis('a=a0,a1')
>>> b = Axis('b=10,15')
>>> c = Axis('c=1..3')
>>> arr = ndtest((a, b, c))
>>> idx = full((c, b), 2)
>>> arr[idx]
 a  b\c   1   2   3
a0   10   1   1   1
a0   15   4   4   4
a1   10   7   7   7
a1   15  10  10  10
>>> a = Axis('a=a0,a1').astype(object)
>>> arr = ndtest((a, b, c))
>>> arr[idx]
TypeError: unhashable type: 'numpy.ndarray'
>>> arr[X.c[idx]]
 a  b\c   1   2   3
a0   10   1   1   1
a0   15   4   4   4
a1   10   7   7   7
a1   15  10  10  10