Get a better way of looking at the gaps in the timeseries pre-WPE calculation

[colebrookson]

Right now it's just raw at all the sites, so let's get:

1. A plot of average gaps (maybe with average size, how many avoided, etc) for each challenge across all the sites across the different rolling window approaches. Maybe paneled by variable type?
2. A plot that does something to show compare the different gap-closing approaches across all the different challenges
3. Something that shows, across the challenges, what kind of data we're actually looking at from all the different time series