Mike, I believe my line of thought on this to be the same as yours- which is that if the bicycle tire leaves an imprint in the path surface, the surface is probably to soft to be accurately measured by a bicycle calibrated on a hard surface.
On the course I measured last year where the second bicycle and mine barely agreed on the crushed limestone sections, the path was on a former rail bed, but there was a failry steep slope on both sides. The curves were long and gradual so we didn't move side to side much, but exactly where the edge of the runnuing surface was could be subject to interpretation.
That being said, I can't say I disagree with Pete. Even if we are not as accurate as we might be under other circumstances, how else would these courses get measured? Steel tape? Doubtful anyone would make the effort unless it was a championship course of some kind. Bike computer? Not calibrated as well as we calibrate- and subject to the same issues regarding surface. GPS? Probably still not as accurate.
If our mission is to ensure runners are ruinning on couses that are as accurately measured as practicable, in most cases I don't think we're betraying ourselves by measuring on crushed gravel with a bicycle calibrated according to USATF procedures.