Fix precision drop when indexing a datetime64 arrays.

[fujiisoup]

• Closes Not able to slice dataset using its own coordinate value #1932
• Tests added
• Tests passed
• Fully documented, including whats-new.rst for all changes and api.rst for new API

This precision drop was caused when converting pd.Timestamp to np.array

In [7]: ts = pd.Timestamp(np.datetime64('2018-02-12 06:59:59.999986560'))
In [11]: np.asarray(ts, 'datetime64[ns]')
Out[11]: array('2018-02-12T06:59:59.999986000', dtype='datetime64[ns]')

We need to call to_datetime64 explicitly.

[shoyer]

Were you able to identify the underlying issue? This feels like something that might likely be a bug in pandas or numpy.

[fujiisoup]

It looks that numpy fails to cast pd.Timestamp to np.datetime64[ns].
But I am not sure it is an expected behavior.

I added a link to numpy's issue.

[shoyer]

Thanks for adding the issue link. This makes sense to me now.