Record some test databases, and build tests for sample processing

[sobri909]

Record some test databases, and build tests for sample processing

[sobri909]

Another possibility is to also save the raw CLLocations and CoreMotion input objects to the database, purely for testing purposes.

Aside: Storing the raw input data during normal operation would be a bad idea, because even just a single hour of recording can include tens of thousands of raw data objects (for example the accelerometer data at 5Hz).

Building testing infrastructure for replaying that raw data would be nightmarish, and would require creating an unrealistic environment in the process. So I don't plan to do that. But having a record of that raw data would help with debugging various low level issues, to determine why the ActivityBrain made various decisions in various cases.

A case that comes to mind is underground train trips. Raw data is extremely sparse during underground train trips, to the point where there is usually little to no location data available. The subtle differences in how the CoreMotion data is interpreted in context of the lack of location data can create unexpected and undesirable results.

Being able to pick through the exact raw data for the moment where an incorrect decision was made on such a trip would be very helpful.