If it were poor electronics design that would show up in SNR tests, but, it doesn't. The 5D Mk III outscores the D800 in that and almost every area except DxOMark tested DR.
No it would not. SNR test is carried out at middle gray where the signal is not affected so much by read noise.
LTRLI is correct
SNR can be measured at various levels and they're not using the dark end of the range or you'd REALLY see the results on a graph!
SNR of Canon's output is notoriously poor at the dark end where read-noise is a significant fraction of the signal level and that's where and why it only shows up in shadow areas.
That's also why total dynamic range results come out as they are.
Canon typically does a little better with highlight headroom but they lose considerable EV range at the dark end where SNR is the limiting factor to what constitutes allowable measurement range.
Because the Exmor sensor has a much lower read noise it can produce acceptable SNR at darker levels than the Canon system by a couple EV or more which adds to the total DR result.
You're free to believe other raw converters can't properly interpret the CR2 file but I believe you're quite wrong about that.
And if that doesn't convince you, just look at real world results of images shot with the latest Nikon sensors vs. Canons.
I have plenty of images from my 5D2 where noise is visible in shadow areas with only +1 EV of push applied, even using DPP. Add +2 EV push and you need to start using noise reduction to get rid of the mess.
You just do not have this low ISO shadow noise issue with the recent Nikons!
After buying and testing my first Nikon body, and comparing it to about 15 Canon DSLRs I've been using over the past 5 years, while using DPP, ACR(LR & PS) and DxO to process my raw files, I DID NOT discover that Canon's performed any better than Nikon in this one critical area. In fact, quite the opposite is true. A sub-$600 Nikon body totally blows away every Canon I've ever used in dark shadow detail retention and lack of noise. And that includes the otherwise well-regarded 5D2, 7D, 60D, 40D, etc. And I'll put it up against the new 5D3 at low ISO as well.
The D5100 performed so well I ordered a D800, a D800e and another D5100.
Now I'm sure you don't know my technical background but do you really think I'd spend $10k on new Nikon gear to get better DR performance by mistake?
For another angle on this, Canon's G11 and G12 use, AFAIK, SONY sensors in them. They also produce CR2 raw files.
Wanna have a look at their low ISO raw dark noise compared to Canon's EOS DSLRs using Canon's CMOS sensors?...
Try explain that one.