I address your questions one at a time:

1). Does anyone actually rely on CFFT as a general measure of cortical processing or intelligence?

The short answer is yes. I do not know how many people in real world settings use CFFT as a diagnostic measure, but as indicated in the introduction to this paper, there are many examples in the scientific literature that point to CFFT as a diagnostic measure. In fact, much of my motivation for writing this manuscript was after having read in a number of papers, particularly in psychopharmacology journals, statements claiming CFFT to be a measure of total information processing capacity (c.f. ref 11), typically without citation. As far as I can tell this dates back to the 1940s and 1950s, and is inspired by papers such as that by Halstead (ref. 6) and Tanner (ref. 5).

2). but doesn't performance on nearly all such tests benefit from training anyways?

I am not sure the point of this question. First, CFFT typically does not benefit from training (and didn't change in most of the experiment presented in this paper). Second, in the main experimental group learning was found for CFFT and the paired motion direction, but not for other motion directions. This is taken as evidence that learning is not a general improvement that transfer to all skills, but instead a highly specific improvement to a couple visual skills.