I2C pull interface and proper averages

[LorenzMeier]

We should be resetting the integrated flow to zero on I2C reads so the reader does not loose samples, independent of his readout speed.

[priseborough]

The average flow rate in mrad/sec could be sent across using an int16, as could the averaged x,y,z gyro readings. 1 mrad/sec is enough resolution for our purposes.

The averaged flow readings should not be motion compensated.

[LorenzMeier]

@dominiho We discussed a double buffer for flow and gyro values so that each I2C read command gets the exact difference since the last read - is this now fully implemented and this issue can be closed?