As for the approach to testing:
Keep in mind, Nikon's marketing guys adjust ISO scores to deceive amateur customers.
On the top end, the
actual sensitivity is about half a stop worse than advertised, so when comparing noise results you must not compare in M mode with equal setting, but adjust EV +0,5 on Nikon (e.g. longer shutter).
Nikon D810 vs Pentax K-5 II vs Nikon D800E About sensor technology levels:
Also keep in mind that the noise performance of the
D810 sensor is exactly the same as the K-5 II sensor performance:
Nikon D810 vs Pentax K-5 II vs Nikon D800E. It's just larger. So when pixelpeeping at 100% you simply have to expect K-5 II level results, unless it's a brandnew sensor after the D810.
Minimum noise levels:
What's also worth noting is that the D800E sensor reaches the same levels of minimum noise performance as the D810, but earlier, with lower ISO / faster shutter speeds, which is clearly
better. If you want minimum noise on a D810 you need to use one stop faster glass or twice the shutter speeds, since you are forced to go down in ISO settings.
Dynamic range also same on D810 sensor tech level as K-5 II: Nikon D810 vs Pentax K-5 II vs Nikon D800E
And finally keep in mind, that the
dynamic range on sensor is the same between the K-5 II and the D810, better than the D800E. And ISO 70 until ISO 1200 Pentax
K-5 II implementation is even better than D810 by 1/3 stop.
So whatever Nikon fanboys try to tell you that the D810 sensor itself is a significant hardware difference to the D800E should first prove (and that is beyond the "I read" or "I heard" myths or self-important blogger opinions) it with proven facts.
DxOmarks is not really a reliable source, but according to what they publish the K-5 II would then be a Sony sensor engineered to Pentax specifications as much as as the D810 mythically being a different variant of the D800E one.

The K-5 II sensor differences and advantages versus D800E/D810 sensor are of the same (un-) significance.
From the facts you can just deduce there are tiny differences between the results measured and it's not a far call to say that the winning recipe is the firmware driven
processing alone.
Let's have a larger version of the K-5 II sensor with Pentax very good processing and we'll all be fine.
By the way:
There is also the
measuring margin of error. Just look at two times the
exact same sensor done by the same maker:
Sony Cyber-shot DSC-RX1R II vs Sony A7R II
Still some deviations exist up to 0.8db on noise. Go explain this, Ken Dumbwells of the world.