DATE_ORDER not always honored.

Thanks for an excellent library!

Version: 0.7.6

Steps to reproduce:

import dateparser as dp
p1=dp.DateDataParser(settings={'DATE_ORDER': 'MDY','PREFER_LANGUAGE_DATE_ORDER':False})
p2=dp.DateDataParser(settings={'DATE_ORDER': 'DMY','PREFER_LANGUAGE_DATE_ORDER':False})

#good behavior
p1.get_date_data('02-23-2004')
{'date_obj': datetime.datetime(2004, 2, 23, 0, 0),
 'period': 'day',
 'locale': 'en'}

p2.get_date_data('02-23-2004')
{'date_obj': None, 'period': 'day', 'locale': None}

#bad behavior
p1.get_date_data('02-23-04')
{'date_obj': datetime.datetime(2004, 2, 23, 0, 0),
 'period': 'day',
 'locale': 'en'}

p2.get_date_data('02-23-04')
{'date_obj': datetime.datetime(2023, 4, 2, 0, 0),
 'period': 'day',
 'locale': 'en'}

My understanding is that with settings={'DATE_ORDER': 'DMY','PREFER_LANGUAGE_DATE_ORDER':False}, the parser should return None if it cannot parse with the specified DATE_ORDER.

The reason I need the parser to honor the DATE_ORDER (or fail the parse otherwise) is I have documents where the parse format is unknown, but it is known that all the dates in that document have the same parse format. If the DATE_ORDER was honored, I could parse all the dates with different DATE_ORDERs, and pick the DATE_ORDER which is consistent with all the dates in the document.

Start by reproducing the two DateDataParser.get_date_data cases from the issue with DATE_ORDER set to MDY and DMY and PREFER_LANGUAGE_DATE_ORDER disabled. Trace the DATE_ORDER handling for two-digit years; done means DMY does not parse 02-23-04 while MDY still returns the shown date, with a regression test covering both cases.