Thinking about it more, there is no way they drove the 75D or the Bolt to exhaustion. That's because whatever circuit they chose has to work with shorter range BEVs in order to provide the consistency they are trying to get but failing miserably. It is actually very difficult to to proper range testing on real streets... variations of weather, traffic patterns, and so forth can have significant effects. That's why the EPA testing is done on a dyno and synthetically corrected for additional factors like aerodynamics. That way the tests are repeatable. The other way is to run it on closed loop circuits doing laps like Idaho National Labs does and record steady state speeds.
CR's testing on surface streets is subject to a lot of driving style variations. As it turns out, I had to make a ~60 mile round trip last night and so I tried with regen set to normal and set to low. The difference in regen effect is significant as it plays with the Wh/mi average. With low regen, there are fewer really big negative numbers to average into the range calculations. At highway speeds, normal regen can do 60 kW. In low regen, the max is 30 kW and low enough that the brake lights never kick in. At lower speeds, the low regen is almost imperceptible and friction brakes are used all the time. That messes greatly with the Wh/mi average, especially if they are aiming for "normal" driving style and not hyper-miling as they stated. That easily pushes the Wh/mi estimation well over 300 Wh/mi.
I suspect CR used something like a 50 mile test loop, something they can do repeatedly with all kinds of BEVs. They then use the consumption as reported by the vehicle in order to estimate the total range. It isn't just the city driving, what is really important for that range estimator is the regen off the highway. Let say the 50 mile test loop is 40 miles of highway... which is pretty close to what I drove last night (26 out of 29 miles is highway). The Wh/mi while on the highway was over 300 Wh/mi, but then with full regen, it dropped significantly. Without full regen, it didn't drop nearly as much. If they drove the car to exhaustion, the difference wouldn't be that big... but without doing that, the acceleration onto the highway isn't balanced by the regen on the other side so the Wh/mi is too high. Of course, the further you drive, the less of an effect this has, but one has to drive a lot for the average to balance out.
Therefore, this change alone can account for 15-20% difference, especially if they didn't drive the vehicles to immobilization. The rest of the difference can easily be highway versus city driving. Again, the Bolt is more efficient at lower speeds and acceleration/regen, but the Model S is more efficient at higher speeds.
Again, the real test is taking both on DCFC trip jumps as that is when range really matters. Being able to drive 230 or 250 miles at 0-40 mph in city/suburban traffic is not particularly interesting. For example, simulate driving from D.C. to Orlando, or LA to SF, both very common driving patterns that require back to back charging at DCFC EVSE's and driving at normal traffic speeds. That's 65 mph to 75 mph speed limits most of the time. I would choose testing at 75 mph, because the speed limit is often that fast in many places and traffic typically flows even faster than that.