Without being there it is difficult to say, but in general:
For the timing, there are some possibilities. That could be the measurement error of the Witschi machine, I have never seen the specs of a Witschi, so I don't know. Also any real oscillator such as the hairspring/balance wheel setup in a watch does not oscillate perfectly with every oscillation being identical. As a statistical process the exact oscillation periods will form a Gaussien bell curve when plotted.
But there is a simpler explanation: at 17997 the watch is going 14.4 seconds slow, 18002 is 9.6 seconds fast a day.
Is that the difference between the fastest position and the slowest position of the watch?
The amplitude does seem to be very high. Was the lift angle set correctly?
Don