Alan, I wasn't aware of a reading difficulty with the graphs because on my PA272W display I can moderately magnify the material on the website and read everything clearly enough; however I can provide you what will be perfectly readable graphs off line if you send me a PM advising which.
I don't think the use of Argyll would make any significant difference to anything I am reporting here. When I test my own profile (made and tested with i1Pro2/i1Profiler) the average dE was 0.7, which is pretty darn good. The high dE readings we are seeing here is also for me a unique outcome that I have not encountered nearly to this extent with the many paper/profile combinations I've been testing.
As for how we print ordinary photographs, yes I agree - making prints out of LR one would be using Relative or Perceptual Rendering Intent and Lightroom handles BPC under the hood. One can replicate this exactly in Photoshop. LR has no Absolute Rendering Intent, which one needs for testing accuracy, so that is why I don't test out of LR.
We have two items at play determining Black shading appearance: the Maximum Black and the shape of the tonal rendition above it. The article demonstrates that the type of profile affects the shape of the quarter-tone rendition down to the maximum Black that the paper/printer combination can reflect. An M3 profile provides for more tonal separation in the quarter-tone range of a matte paper than we get from conditions M0/M1/M2. Turning to the Maximum Black point, if you examine Figure SB2 in the sidebar, you will see that Maximum Black only deviates from c. L*20 in the condition where one measures the Black patch with a spectro and software wherein M3 is enabled. In that one condition, the statistical value of Black is down at L*1. However, when you make a print with that profile and look at it, the printed Black looks more like L*20. The two lower panels of Figure SB2 are in Relative Intent with BPC, so they would predict outcomes printing from Lightroom with Relative Intent selected. Maximum Black should look similar whether using an M0 or M3 profile, but with somewhat more distinct tonal separation above that using the M3 profile. Perhaps we are saying the same thing! :-)