Add origin parameter to Timestamp/to_datetime epoch support.

[jtkiley]

When using SAS or Stata data, dates are represented as the number of days since 1/1/1960, and other statistical software uses different origin dates. With that in mind, it would be nice to have an origin date that can be specified. See also, #3969.

It's a relatively simple thing, and not hard to work around, of course. However, I end up dealing with date formatting on just about every data set I import, and I imagine that lots of others do, too.

Currently, I do something like this:

import pandas as pd
import datetime

EPOCH1960 = datetime.date(1970, 1, 1) - datetime.date(1960, 1, 1)

data = pd.read_stata('./data.dta')
data['date'] = pd.to_datetime(data['date'],unit='D') - EPOCH1960

In R, the as.Date() function takes an origin parameter for numeric types (see, manual). So, in R, the date part would simply be:

data$date <- as.Date(data$date, origin = '1960-01-01')

[jreback]

sure this could be a parameter to to_datetime (only) as that is the datetime converter
could work alongside the unit parameter (like #11276).

In fact that should be origin rather than unit now that I think about it. xref #11470