That implementation of the max(A, X) operator that is described in the article is very clever. Obviously not super practical, since it needs to allocate one specific zeropage location for it to work ($8A), but that type of requirement is pretty typical for lovely tricks like this one! And the beauty here is in the trick, not whether or not there may be some practical application of it.
But this also raises the question: is there some clever way to use the undocumented SAX instruction, which does an AND between the A and the X registers, to achieve something similar? There is and old trick to compute min and max using no branches, which is not immediately applicable (https://graphics.stanford.edu/~seander/bithacks.html#Integer...) - but maybe there is some other trick hiding in there?