The author claims that the rules described, "extending" the standard C++, are enforcing memory corruption, and it is this author described subset that is still unsafe, not C++ in general. Think about english vs. americanized english, largely the same, but two distinct entities.
In C++, you are free to shoot yourself in the foot. In Rust, you have a Government inspector ensure that you always point the gun not just in a "safe" direction, but only toward a crosshair target at a designated gun range.
You're probably right, it's probably more of a psychological issue than an issue of randomness. Perhaps his users simply trust dice more than they do "atmospheric noise with a certification of randomness across sliding windows", even though minor imperfections in dice (and dug-out pips are more than minor imperfections) will cause them to be much less fair.
That said, it would be really interesting to see the referenced thesis-like papers. Or use high quality Vegas dice for awhile and compare the outcome of those to the purely random data.
Oops.