The beautifully simple thing about the drilled solution is that the 3.5mm VR sensor will always be an exact distance from the un-drilled surface of the flywheel, to all intents and purposes, parallel. Add a flat nut to that equation vs the curvature of flywheel and the sensor to nut relationship will be variable. If the hole for the first solution isn't drilled at an exact angle it doesn't really matter as the depth of hole will not really be an issue, whereas any imprecision in the angle of the drill hole to the nut will add an angle to the surface of the nut, further exaggerating that lack of parallel surfaces. Having just experimented with a 5 pence piece (the closest to a nut without a shaft I could find), there is about 1mm difference at the edges to centre height. That's a lot of variability in pick up each rev cycle.
I'm not sure I agree about the additional balancing being required as 3" - the difference in balance one side to the other is a 10mm wide hole undrilled to a width of 5mm.
Having measured the location of the clutch bolt holes, the depth of metal remaining on the outer perimeter when drilled to 5mm would still be greater than the depth of metal from the hole to the inner edge. I don't think this is a problem compared to ensuring deeper drilled holes exactly miss either the three through holes or the clutch bolt holes, given you have to distribute both sets of holes evenly around the surface anyway.