Recreate Apr.22.Publication.Summary.Data with longitudinal data

[lboel]

I started working in the main Looker Report and noticed that were is at least one case where a patient does both self-mixed and basal-bolus.

This deeply relies on the implementation off #76 but having both in one cell ofa patient makes it a little trickier than I previously thought.

I think we implement it first with a forced choice (in #76) and then we kick them out afterwards if the cases there both insulin regimens are used have to be filtered out (after all I think this shall be used to compare the performance of both ways to treat Db1 other the day, so mixed cases need to be filtered out anyways.)