raise ValueError in DataFrame.groupby.apply when index is duplicated

Exception is raised when the index is duplicated

Fro example:

```
df = pd.DataFrame([['a',1],['b',1],['a',2],['b',2]],columns=['c1','c2'], index=list('ccdd'))

df.groupby('c1').apply(lambda x:x)

raise ValueError("cannot reindex from a duplicate axis")
ValueError: cannot reindex from a duplicate axis
```

This is a new issue affecting 0.17
