I had a look at your test shots for 70, 135 and 200mm. What I believe I am seeing is that there is always one corner (or one side) that is definitely softer than the others (lower right mostly, upper right @200mm) but not catastrophically so in each case. This corner then remains visibly worse at f4 and eventually sharpens up at f5.6.
That could indicate some decentering, but could also still be a problem with the test setup. What also made me wonder is the distortion @200mm... the lower border of the test chart shows more pincussion distortion than the other, although the chart seems to be a bit shifted to the top. But then, shouldn't the upper border (being closer to the edge of the frame) show more pincussion distortion?
Point is: your setup isn't perfect (and I doubt it can be with conventional means). Maybe you should check for decentering with a different approach that is less sensitive to test setup, focus issues and shallow DOF: Go outside on a clear day with good light, find far away target (i.e. far away for infinity focus) with good contrast (lone tree in the distance, a tower, tall building, church etc.). Put that into the center of the frame and assure perfect focus (lifeview MF or AF), take test shots at relevant apertures, maybe a few shots for each aperture so that you can rule out camera shake (or better yet, use a tripod ;-)). Now, without changing focus, reframe to put the "test object" in one frame corner, again doing test shots for each aperture. Repeat that for all four corners, making sure that in each shot the test object is roughly the same distance from the image corner.
That should give you five sets of images, one for the center, one for each corner, at each aperture. If you made several shots for each area and aperture, pick the best. Now create a "collage" of the shots of each aperture so that you can visually compare them on one image. The corner shots will likely be darker and a tad less sharp than the center shot, but they should all be roughly equal. If they are not, this is likely decentering.
I couldn't find a good site explaining the procedure in English, but I found this page in German... I think the pictures are pretty self-explantory:
http://www.gletscherbruch.de/foto/test/dezentrierung/dezentrierung.html