ENH: Allow support for dtype datetime64[ms] or other units or numpy.datetime64 cols.

[isichei]

Is your feature request related to a problem?

I would like to be able to use Pandas datetime datatype column for datetimes beyond the bounds for ns units (which are 1677-09-21 00:12:43.145225 and 2262-04-11 23:47:16.854775807).

As will often be dealing with dates above and below these thresholds. Apologies if already open on another issue but couldn't quite find this. Not sure if tied in with this issue

Describe the solution you'd like

df = pd.DataFrame({"col": ["2020-01-01", "2999-01-01", "1990-01-01", None]})

One of the following:

import pandas as pd

df = pd.DataFrame({"col": ["2020-01-01", "2999-01-01", "1990-01-01", None]})
df.col = pd.to_datetime(df.col, base_unit="ms")
df.col.dtype # datetime64[ms]

or

import pandas as pd
import numpy as np

df = pd.DataFrame()
df["col"] = np.array(["2020-01-01", "2999-01-01", "1990-01-01", None], dtype='datetime64[ms]')
df.col.dtype # datetime64[ms]

API breaking implications

Not sure, I assume this might be a big ask depending on how the column casting internals work.

Describe alternatives you've considered

At the moment the only workaround is having an object column with datetimes. However I believe this stops you from using the .dt method on the column which is really handy.

[mroeschke]

Thanks. This falls into the same core request in #7307, so going to close of a duplicate of that request