Mark, if I understand your data, it seems to bolster Bob Thurston's contention that calibrating on pavement typically if not always results in a slightly long measurement on grass. If this is accurate, it seems to me there could be a way we could establish criteria for a new category of course measurement.
Instead of issuing a certificate, we could issue an informal "letter of accuracy" or something similar. The criteria would actually be stricter than for road courses, since the measurement path would have to be painstakingly defined and documented.
Current GPS instruments record single points with good accuracy when there is clear sky view. With sufficient points identified, it seems to me the measured course could be accurately recreated for race day. Stream hops and blow downs could be measured over with a steel tape with no loss of accuracy.
I have measured a few XC courses in which I took great pains to document dozens of points to define the measured path. I would be willing to wager that all of these measurements could stand up to a rigorous tape pull. In fact, I would be surprised if they aren't all at least a few yards long.