BUG: multiindex slicers requires a unique index

[jreback]

This is actually not required, and so should work I think
(requires lexsorting only)

In [70]: df = DataFrame(dict(A = ['foo','foo','foo','foo'], B = ['a','a','a','a'], C = [1,2,1,3], D = [1,2,3,4])).set_index(['A','B','C']).sortlevel()

In [71]: df
Out[71]: 
         D
A   B C   
foo a 1  1
      1  3
      2  2
      3  4

[4 rows x 1 columns]

In [72]: df.xs(1,level=2)
Out[72]: 
       D
A   B   
foo a  1
    a  3

[2 rows x 1 columns]

In [73]: df.loc[(slice(None),slice(None),1),:]
ValueError: MultiIndex Slicing requires a unique index