BUG: selection from unordered time-series has incorrect / odd behavior

[jreback]

original mods made < 0.11 in #3136 / #2437

incorrect selections, depending on the start:end pos gettting weird values

also should be able to select a value that's bigger/smaller than the contained range and have it work (this is how ordered series works)

[21]: s = Series(randn(10),date_range('20130101',periods=10))

In [22]: x = s.index.values

In [23]: random.shuffle(x)

In [24]: s.index = x

In [25]: s
Out[25]: 
2013-01-09    0.911882
2013-01-06   -0.310166
2013-01-10    1.090196
2013-01-07   -0.846249
2013-01-02   -0.205013
2013-01-08   -1.284152
2013-01-05    0.018700
2013-01-04   -0.503527
2013-01-01    0.073035
2013-01-03    0.695904
dtype: float64

In [26]: s['20130104':'20130105']
Out[26]: Series([], dtype: float64)

In [27]: s['20130104':'20130107']
Out[27]: Series([], dtype: float64)

In [28]: s['20130102':'20130107']
Out[28]: Series([], dtype: float64)

In [29]: s['20130102':'20130104']
Out[29]: 
2013-01-02   -0.205013
2013-01-08   -1.284152
2013-01-05    0.018700
2013-01-04   -0.503527
dtype: float64

[jreback]

closed by #3452

Revised using code from above

In [14]: s
Out[14]: 
2013-01-01   -0.950737
2013-01-06    0.766329
2013-01-02    0.234236
2013-01-07   -1.009092
2013-01-03    0.519744
2013-01-09    0.736012
2013-01-04    1.063657
2013-01-05   -0.383949
2013-01-08   -2.117098
2013-01-10   -0.761778
dtype: float64

In [15]: s['20130104':'20130105']
Out[15]: 
2013-01-04    1.063657
2013-01-05   -0.383949
dtype: float64

Ordering is whatever it is currently (no sorting done on return index)

In [16]: s['20130104':'20130107']
Out[16]: 
2013-01-06    0.766329
2013-01-07   -1.009092
2013-01-04    1.063657
2013-01-05   -0.383949
dtype: float64

In [17]: s['20130102':'20130107']
Out[17]: 
2013-01-06    0.766329
2013-01-02    0.234236
2013-01-07   -1.009092
2013-01-03    0.519744
2013-01-04    1.063657
2013-01-05   -0.383949
dtype: float64

This is out of range selection!

In [18]: s['20130102':'20130204']
Out[18]: 
2013-01-06    0.766329
2013-01-02    0.234236
2013-01-07   -1.009092
2013-01-03    0.519744
2013-01-09    0.736012
2013-01-04    1.063657
2013-01-05   -0.383949
2013-01-08   -2.117098
2013-01-10   -0.761778
dtype: float64