I also talked to NAD about this.
They indicated that the dynamic power performance isn't a good test in their view (I'm paraphrasing) and that NAD has long intentionally designed their amplifiers to provide similar power into both 8 and 4 ohm loads as a means to provide similar low distortion into both loads. (I will share more once I get permission).
I won't share anything specific to NAD yet, but...I will say this. The modules they license and supplies they use CAN be designed intentionally in a way that will make it so that their power won't change much with load. This would be an intentional and desirable trait. An overly simplistic way of describing this is that the supplies are far more regulated than they once were and it is possible to lower the rail voltage (on purpose) to ensure it can be better regulated into lower impedance loads. THX would ding an amp that behaves this way, but this is exactly how I would design a modern amplifier. I did something similar with my own last amplifier, the switching supply operates at a different mean rail voltage as the impedance drops. It can range anywhere from 60 volts up to 72 volts. With a solid flat 8ohm or higher load, it will operate at the 72-volt rail. If the impedance is more typical of 8ohm speakers or is, in fact, a 4-ohm speaker, the rail drops down, typically to around 65-68 volts and remains at that voltage under a very wide range of loads. Much like unregulated linear supplies, the power supply stops strictly regulating its rail voltage at peak current draw and allows the voltage to drop as low as 60 volts. At that point, it hits a hard current limit and actually shuts down momentarily to protect itself. I didn't design the supply just specified its operating parameters.
I also mentioned to James that I thought there speaker measurements had some issues. He noted the focus on power response, which I agree is a strange focus. I think it makes sense to look at the smoothness of the response as you move around the speaker, looking for little major variation. However very different speaker designs will have very different power responses, and neither is necessarily wrong. It seems surprising they didn't include directivity index instead, and I would simply have made the smoothness of the DI the focus, rather than a particular level or even tilt. Some designers like a tilt in the DI. Some want it at 0 and some want it at 20db's (Speakers with lasers on their head). It's all a design decision. The peak SPL test is one I'm heartened to see included. James and I disagreed on the merit of such a test, but...I have to say, I wish they would describe in more detail what they are doing in that test. Those values seem much higher than I would have expected as possible from the speakers under test.
The last point, I think a lot of their tests are of questionable merit. I would much prefer them to ensure that the tests either a) have a strong association with perceived sound quality, or b) have a reasonable engineering purpose. The THD tests aren't ideal, you could have an amp with fairly low THD that still has audible distortion. While not a well studied topics, certainly enough papers exist dismissing the THD spec. My mentor, Earl Geddes did a nice start toward this, noting that certain combinations of harmonics can lead to a low absolute value and audible distortion. The dynamic power test seems silly, increasing power into lower loads doesn't benefit sound nor is it a sign of good engineering (in the modern era).
