Consider including time series completion code in vignettes

[brookslogan]

E.g., in archive.Rmd, there is a plot that, for some data sources or circumstances, might have gaps that are plotted improperly by geom_line (but don't appear to have gaps in this particular case, when the vignettes are run today). Consider using the group_by complete full_seq pattern, if nothing else but to demonstrate it for users. It's not only archive.Rmd needing potential changes; we'll need to determine where all such a change belongs.

[brookslogan]

May also want to show a full-key complete pattern, and show difference vs. group_by complete.