What we know should suspect (absent more data on test standards) is that these numbers may be so susceptible to manipulation by unscrupulous manufactures (of both passively and actively thermally managed packs) that these numbers may be quite misleading.
True. Although let me point out that the farther you diverge from the simulated norms of any of those test cycles (which are meant to mimic average behavior), the less accurate they become. The hypermiling example is a good case in point. It would appear that you are upset that the predicted range/capacity during the hypermiling test did not live up to the extrapolated estimate that the EPA test would have predicted, when actually there is a perfectly reasonable explanation. Nothing nefarious took place, just a higher than expected overhead due to the cooling system.
While it's a true statement that the numbers are "misleading" if you are using them to rely on predicting range/capacity for a highly atypical usage pattern (such as a hypermile run), but I would argue that it's more of a misuse of that data, which carries the disclaimer right on the label itself, that:
Actual results with vary for many reasons, including driving conditions and how you drive and maintain your vehicle.
As you said, it is up to the end user to determine their own actual range for their own vehicle and not to rely on the Monroney sticker. But you don't have to assume that the numbers of being manipulated (although it is good to have a healthy skepticism). In Tesla's case, under NORMAL driving patterns, for example, the numbers were manipulated, but the consensus is that Tesla actually sandbagged the numbers (to keep a separation between the Model 3 and Model S) and that reported ranges actually exceed
the Monroney value.