PERF: tz_localize(None) from dst, asvs

[jbrockmendel]

This makes tz_convert_dst follow the same pattern we use elsewhere, and brings a perf improvement along with it.

In [2]: dti = pd.date_range("2016-01-01", periods=10000, freq="S", tz="US/Pacific")                                                                                                                          

In [3]: %timeit dti.tz_localize(None)                                                                                                                                                                        
89.3 µs ± 627 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)  # <-- PR
100 µs ± 6.36 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)  # <-- master

The boost comes from not doing a copy, so makes a bigger difference in the scalar case than the array case.

In [5]: ts = dti[0]                                                                                                                                                                                          

In [6]: %timeit ts.tz_localize(None)                                                                                                                                                                         
12.4 µs ± 188 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)   # <-- PR
16.1 µs ± 215 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)   # <-- master

The improvement is much bigger for fixed offsets:

In [9]: tz = pytz.FixedOffset(-60)                                                                                                                                                                           
In [10]: dti2 = dti.tz_convert(tz)                                                                                                                                                                           

In [11]: %timeit dti2.tz_localize(None)                                                                                                                                                                      
34.8 µs ± 1.09 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)  # <-- PR
71.8 µs ± 2.99 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)  # <-- master