always use object as dtype when creating a string series

[clembou]

The behaviour of

In [3]: object_data = ['silly walks', np.nan, 'dead parrot']
In [5]: pd.Series(object_data, dtype=str)

as changed in pandas 0.23.
it used to return:

Out[5]:
0    silly walks
1            NaN
2    dead parrot
dtype: object

but now returns

Out[5]:
0    silly walks
1            nan
2    dead parrot
dtype: object

using str as a dtype was probably a bad idea in the first place, so I changed it to object which is a supported dtype in pandas, and behaves as expected.