Interpolation is not working correctly for small study areas and/or short interpolation windows

[cgmorton]

The current interpolation approach will not return all "expected" images when the interpolation window is small or the study area is really small (e.g. for a single path/row). This seems to be a result of the saveAll join dropping the "source" images when there are no "target" images to join with. What you will see is that for a small interpolation window, the only interpolated images that get included are those between the Landsat target images/dates when there is an image both before and after, but not the image for the actual Landsat dates.

This probably isn't that big a problem for large study areas with the default window size (32 days) when there is likely to be at least one bracketing image somewhere to join in, but it is definitely a bug in the approach.