Potential fix for moving window percentiles

Edited the filter_data_by_time function in utils.py to handle situations with moving windows where one side of the window crosses over the calendar year. I wrote an additional if, elif, else statement that handles these edge cases. One thing to note is that I think leading values and trailing values are added to each day in the dataset, such that if leading_values = 15 and trailing_values = 15, it is actually a 31-day moving window, where 15 days are added on either side of the focal day.

Created to address issue #40 (closed)