BUG: Bug in slicing a multi-index level with an empty-list (GH8737)

[jreback]

closes #8737

In [5]: multi_index = pd.MultiIndex.from_product((['foo', 'bar', 'baz'], ['alpha', 'beta']))

In [6]: df = DataFrame(np.random.randn(5, 6), index=range(5), columns=multi_index).sortlevel(0,axis=1) 

In [7]: df
Out[7]: 
        bar                 baz                 foo          
      alpha      beta     alpha      beta     alpha      beta
0 -0.725086 -0.715498  1.141087 -0.586018  0.906581 -0.260218
1 -1.111794 -0.393149  2.101567  0.648850  0.366892  0.554206
2  1.258239  1.472659 -0.174970 -0.355639 -0.843127  1.101716
3 -0.775790  0.419361 -0.521444  1.245519  0.759791  0.183877
4 -0.001001 -1.526741 -1.721501  0.527531  0.407099 -0.415245

In [8]: df.loc[:, ([], slice(None))] 
Out[8]: 
Empty DataFrame
Columns: []
Index: [0, 1, 2, 3, 4]

In [9]: df.loc[:, ([], slice(None))].columns
Out[9]: 
MultiIndex(levels=[[u'bar', u'baz', u'foo'], [u'alpha', u'beta']],
           labels=[[], []])

[jreback]

merged via 1887ba5