How to pass holiday lower_window /upper_window for weekly aggregated data in prophet time series modelling

Question

Exploring the fb prophet package, I'm below code for passing holidays to daily data time series.

superbowls = pd.DataFrame({
  'holiday': 'superbowl',
  'ds': pd.to_datetime(['2010-02-07', '2014-02-02', '2016-02-07']),
  'lower_window': -1,
  'upper_window': 1,
})

In the above code, my lower_window and upper_window is 1 day each.

But if my data becomes weekly aggregated, i understand my holiday date needs to fall on particular date, so as to get factored-in. But what should be the lower and upper window to pass in this case.

Since the data is weekly aggregated, does it make sense to pass lower_window parameter still in terms of days