Because the JPEG tone curve (measured by DPR) tells us nothing about the DR captured by the sensor. It really just indicates which jpeg settings produce low contrast images and which jpeg settings produce high contrast images. In some recent reviews DPR has tried to clarify the difference, but unfortunately they continue to use the term Dynamic Range on their tone curve test.
The Shadow Noise test on page 26 of that same review tells us more about the Dynamic Range of the sensor. You can clearly see the Canon has more noise in the shadows. The D800 can capture more stops of highlight and shadow detail while remaining below a specific level of noise. That is how it achieves a higher DR score.
It is not a conspiracy. Canon just hasn't improved the DR of their sensors yet; I am sure they will catch up eventually.
As a Canon fan, you know there are many aspects to a camera beyond sensor performance (e.g. lens quality, AF performance, metering accuracy). For example, the Canon S100 sensor is better than the Lumix LX5 and Olympus XZ-1, but the Canon S100 lens is so bad that the other 2 cameras are much better options. So, DXO sensor testing only tells us half the story.