I beleive your analysis is right (as usual). LOL
Fundamentally one struggles to measure more accurately than the base resolution of the system one has as the source even if you have any device to decouple the stop start nature of the watch and interpolate. One might imagine a large freewheeled flywheel or something - the problem is you can never really predict the position of that. This is a problem we have in measuring crankshaft position from large resolution (e.g. 60 tooth wheel or 6 deg) flywheel cutouts. The variable nature of compression, combustion and expansion in an engine cycle is enough to give us an error which we don't know or we have to correct for by calibration with a finer device than we need (e.g. a 0.5 degree encoder sometimes 0.1 degree).
It would be fascinating to know how the measured timing compares to a quartz base - whether there is some kind of consistent effect which could be corrected for within known bands. FPJ could even provide those for customers?
At the moment it feels like the watch can only measure to +/- 0.045s! or something based on your calcs. ;-) And hey, that's not bad!
Velociphile