Develop a way to deal with time series data

[justaddcoffee]

Per convo with @realmacin @cmungall @justaddcoffee @wdduncan

We'd like a simple way of modeling time series data, for example these use cases:

• clinical course of patients over time (time element: days from first symptom or days from diagnosis, probably)
• number of infections/death by date and geographical location (time element: probably YYYY/MM/DD, for example here)
• time course data for cells infected in vitro (time element: hours or days post infection, like #12)
• possibly twitter data (#32)

[wdduncan]

Time should be in "YYYY-MM-DDTHH:MM:SS". E.g.: "2007-03-01T13:00:00Z". "Z" is for Zulu. See ISO 8601: https://en.wikipedia.org/wiki/ISO_8601

This also has a section explaining duration formats, which I have not used before.

If possible, days from diagnosis and the like should be calculated. Although, sometimes you don't have enough information to do the calculation. Perhaps we can have a general property like "event duration".