DxOMark's ratings are as subjective as anyone else's- interpretations of measurements, weighing what they think is more important.
But their measurements are nearly always spot-on. They've tested plenty of Sony 16MP and 24MP sensors, and with the 24MP sensor on the K-3, the general lack of real resolving power among Pentax's otherwise very neat primes is a definite drawback. It'd be better if Sigma, Tamron and Tokina had a full lineup in K-mount, but even third-party lens selection is pretty anemic- and precious few of those lenses live up to the full potential of first-party lenses on other systems.
That said, the only 'quality' issue Pentax has is in lens resolution; from what I've seen and heard (and would love to experience first hand) their rendering is stellar, and that makes up more than the difference unless you're printing extremely large. Nikon's new 58/1.4G isn't three times the price of the 50/1.4G for nothing, after all.
I - and many, many other folks - find DxO's lens measurements to be off. /shrug Just because a number is put to something doesn't make it factual.
You can attach a Zeiss or Leica or whatever lens to a Pentax body and it's stabilized lens. That's my "third-party lens selection", among the best in the industry.
On FF vs. APS-C vs. MF- this'll be different across different systems, but note that FF cameras are generally the most capable cameras in a system. It's not all about measurebating either- from personal experience, moving to FF meant being able to shoot well in far lower light, having even greater latitude with post-processing, and having even greater control over depth of field. I can get both shallower depth of field than a crop system is capable of realizing, while also being able to stop my lenses down further without butting up against the effects of diffraction. They're the sweet spot of this kind of shooting, where shallow depth of field requires more care on smaller sensors, and as PenGun mentioned above, diffraction from extremely narrow apertures on medium format and larger systems becomes a working limitation. And then there's the lack of CMOS sensors above 35mm format that keeps such cameras below ISO 1600, when a modern Canon can hit ISO 6400 with ease.
I don't believe I ever mentioned anything negative about FF, nor implied that APS-C is superior. All I said is that most people who want it don't actually need it.
Yes, you can stop down lenses further before hitting diffraction (pending the lens design, of course), but given the shallower DoF inherent in FF, that does not translate to greater DoF. While "DoF control" is desirable, shallow DoF is frequently used as a crutch by people who can't shoot interestingly or creatively. And
it's frequently abused by people not having enough of their subject in precise focus. f/64
Any modern camera can hit ISO 6400 with ease, so I don't see a problem there... except maybe for modernist noobs who freak out at a hint of imager noise.
Since you mention "personal experience", I'll mention that mine includes m4/3, APS-C, FF, MF and LF; the latter two not digital, though.
The point lurking in the background of my posts is that the nature of the gear is... just not that important. The common stuff gets obsessed over and talked to death. I just got an MTO 500/8 mirror lens that's actually 548/8.8. It passed factory inspection in 1970, the certificate filled out by hand and lens resolution measured at 32 lp/mm. Sounds fun...