I don't have anything other than those PlugLoad numbers and reports. I do estimate the 10% Load efficiency when none is reported by comparing the dropoff of similar units, which depending on brand and series runs around 5% to 10% when comparing 10% Load efficiency to 50% Load efficiency. For back of the envelope calculations that's probably accurate enough.
I agree that it would be really helpful from a Total Cost of Ownership standpoint to know what a given PSU efficiency will be when the system is in a low-power idle state. Using a lower-bound like gary suggests should give a worst-case efficiency. I suspect that the simple reality is for unRAID and other systems that don't make use of a discrete GPU, the PSU chosen is overpowered by hundreds of watts. There is some small but measurable savings to be had by sizing your PSU smaller so that it runs at a more efficient load rating, but with power draws so small it almost becomes moot.