Loop RC is dependent on sensor start time

[motinis]

The current LoopMath.decayEffect always rounds down to the nearest 5 minutes. This means that the RC effect can be identical (on the clock, not relative to the time of the latest glucose reading) for glucose readings at 09:00:00 and 09:04:59. This also means that the RC effect difference is "large" between a glucose reading at 09:04:59 and 09:05:00 (where the latter will see the first effect at 09:05 and the latter at 09:10).

This can be fixed by changing the application to a quadratic model which satisfies the equations where time is then taken relative to the latest glucose timestamp.