I think Bernard's efforts are quite interesting. The problem is that I can use his target method to fool a Canon 5D, Canon T3i, Canon G1 X, Fujifilm X-Pro 1, Nikon D700, D5200, D7000 and a D800 into consistently front- or back-focusing. It leaves the impression that the AF in each body needs fine tuning. But when I check the same cameras using the same lenses shooting a known-flat chart that is mounted parallel to the plane of each camera sensor, AF becomes consistently accurate. How much physical bias by the photographer at home doing such testing figures into this sort of DIY effort is another unkonwn factor, but a variable still. With a less-than-optimal parallel/plane test target I can also force AF inaccuracy.
There's no doubt that Bernard's chart method can be accurately used, but it takes so much effort, careful lighting and attention to focus point positioning (in order to get a truly accurate focus lock in the first place) only to almost always confirm the degree AF accuracy that is so much easier and faster to confirm using a parallel/plane testing method.
By the way, sufficient static charge buildup or discharge on a table surface can gradually lift an unsecured paper target sheet or gradually lower it - a few mm in each case - to render all of the AF testing effort worthless.
For a camera that fails the parallel/plane chart focus testing method, AF fine tuning might be needed, in my opinion. The main problem with any sort of AF fine tuning is that the camera owner has to assume that the camera's AF system will always react precisely the same way at each and every press of the AF-On button or shutter button. As many tests published by LensRentals and other testers have shown - particularly in the most extensive sample testing of all the top camera bodies over the past couple of years - AF systems are fundamentally inconsistent in terms of absolute accuracy. Sequences of identical test shots with the same camera and lens results in a range of results which reflect the engineering/design tolerance variables inherent in the AF system. The point is that if we then apply AF fine tuning adjustment to that camera and lens combination, we're not reducing the variability inherent in the design of the AF system, rather merely shifting some median point to another location. But the variability inherent in the AF system remains.
It makes more sense, IMO, to adjust our shooting habits to the particular combination of engineering/design/assembly variation of the particular camera and lens we're using, and to adjust our focus technique over time as we come to understand how the camera and lens react in different shooting situations for different kinds of subjects. That takes time. As we get to know the combo's specific variables, we inevitably refine and improve our use of the particular camera and lens. Only a grossly misadjusted camera and/or lens will stymie such efforts and such gross misadjustments are very rare.
After having gone through somewhere north of 60 Nikon digital SLR bodies (personal use and review use) and around 30 Canon DLSR bodies (personal use and review use), I have only encountered five with a bad AF system (two Nikon, three Canon). The first Nikon one was my original D700 - the AF system simply failed during the first week I had the thing. My dealer exchanged it over the counter for a perfect one which I used for years. The second Nikon AF system problem occured with a D7000. In the second D7000 I had, the AF system seemed to be completely uncalibrated - an obvious error that escaped the factory. The same dealer exchanged it over the counter for a D7000 which turned out to be perfect. I've still got it. Great camera.
The three Canon bodies were a 30D, 60D and a 1D MK III. The 30D was mis-calibrated at the factory. The 60D had a misaligned mirror. The 1D MK III simply wouldn't lock focus accurately without sufficient prayer, sprinkling of magical pixie dust, and repeatedly murmured incantations. The 30D and 60D were review units and were each exchanged by the PR rep for good ones. The 1D MK III was exchanged by my dealer for a unit that had been unpgraded by Canon (a sub-mirror assembly replacement IIRC). Worked perfectly.
I think that, except in exceedingly rare situations in which a camera escapes the factory having been calibrated at or near one extreme of its AF adjustment tolerances and with the owner then using a lens that is at or near a complimentary extreme of its calibration tolerances as well, focus accuracy is more of a photographer adjustment than a technical camera AF system adjustment. I admit that the biggest difference between my results may be based on the fact that I'm a street shooter - tripods just aren't in my photographic vocabulary - and natural adjustment based on familiarity with the camera's variability in different shooting situations may be even more important than it is to photographers who work with a tripod more often than not. A good, sturdy, well adjusted tripod & head at least reduce the impact of some of the variables in any shooting situation. Then again, AF testing at home is always done on a tripod.
I realize that there are photographers who are tweaking every aspect of their camera performance, and I fully respect the effort. I think anybody who spends the time to squeeze every drop of performance out of their photography gear deserves respect. Nonetheless, I think it's unrealistic to expect camera AF systems (read: all AF systems in use today across all makes and models) that struggle with texture detail at 45 degree angles (or other severe angles) in real shooting situations in almost all lighting to produce test results reliably consistent enough to form the basis of AF fine tune adjustments during DIY home testing.
I believe it's possible that Bernard's approach absolutely nailed an AF problem he was experiencing. Every such test method works for somebody, somewhere. But I also believe that duplicating his method to achieve some perceptible improvement in AF accuracy is too inherently problematic to be worthwhile. These cameras are amazing, but their most conceptually and physically difficult aspects to refine are the AF systems - ironically the one thing in particular that we rely on to help produce the photos we like best. AF systems are not simply mechanical affairs, rather combining mechanical systems with optical systems, management software and digital processing. Expecting absolute consistency out of them is unrealistic and, at least at the current state-of-the-art, impossible.