[BUG] cudf.date_range doesn't match pandas for weekly frequencies

[Matt711]

Describe the bug
See repro

Steps/Code to reproduce bug
cudf:

import cudf
cudf.date_range('2016-01-01 01:01:00', periods=5, freq='W', tz=None)

DatetimeIndex(['2016-01-01 01:01:00', '2016-01-08 01:01:00',
               '2016-01-15 01:01:00', '2016-01-22 01:01:00',
               '2016-01-29 01:01:00'],
              dtype='datetime64[ns]', freq='7D')

pandas:

import pandas as pd
pd.date_range('2016-01-01 01:01:00', periods=5, freq='W', tz=None)

DatetimeIndex(['2016-01-03 01:01:00', '2016-01-10 01:01:00',
               '2016-01-17 01:01:00', '2016-01-24 01:01:00',
               '2016-01-31 01:01:00'],
              dtype='datetime64[ns]', freq='W-SUN')

Expected behavior
Match pandas

[mroeschke]

Just noting that cudf weekly frequency is "fixed" at 7 days, while pandas weekly frequency also has a concept of anchoring at a start date (e.g. Sunday in the above example). To fix this would require re-mapping weekly frequencies to a cudf.DateOffset with a day anchoring