I made an observation, less for the K-5 but rather for a D7000 sample shot shown.
Please, refer to
Digital Cameras, Nikon D7000 Digital Camera Test Image for the original image.
It is an IR shot with a distant test chart (sitting under the bell). That's essential as the full test charts are too large to be interesting.
The interesting detail is the little zone plate chart in the center. At this distance, it is fine grain enough to show the zone plate alias pattern.
For a theory of sampling artifacts of the zone plate function ( sin[x^2+y^2] ), please refer to, e.g.,
http://www.cgl.uwaterloo.ca/~csk/papers/kaplan_bridges2005a.pdf
The attachment shows the center crop (200% and heavily resharpened).
You see that the zone plate alias pattern emerges at a 45° angle.
That's astonishing and a possible sign that Nikon samples the raw data in a finer pitch lattice turned 45° rather than the original grid of sensels. That, or the sensor is turned 45° which I would think is highly unlikely
I didn't open the NEF and look into it. Maybe interestig to compare though.
The K-5 images show the alias pattern at the expected 0°/90° angles.
So, I add yet another possible cause for the differences in sharpness: Different levels of sophistication in in-camera demosaicers.