Series[datetime64[ns, tz]].__setitem__ fails to coerce for ndarray value

[TomAugspurger]

What's the expected output for

In [9]: a = pd.Series(pd.DatetimeIndex(['2000', '2000'], tz="UTC"))

In [10]: a.loc[:] = np.array([pd.Timestamp('2000', tz="UTC")] * 2)

On master, we coerce self to object and then set.

In [11]: a
Out[11]:
0    2000-01-01 00:00:00+00:00
1    2000-01-01 00:00:00+00:00
dtype: object

Should we instead coerce value, rather than self?

[jbrockmendel]

We should certainly at least try

[TomAugspurger]

FYI, I have this working on my all-in-one branch (hoping to have a PR tomorrow).

In [4]: a.loc[:] = np.array([pd.Timestamp('2000', tz="UTC")] * 2)

In [5]: a
Out[5]:
0    2000-01-01 00:00:00+00:00
1    2000-01-01 00:00:00+00:00
dtype: datetime64[ns, UTC]