In the implementation, macroscopic fading refers to signal attenuation based on BS-UE separation distance (i.e., pathloss) while shadow fading models the deviations from the average pathloss values due to structural/environmental features such as terrain, buildings and obstacles, as you rightly said. Both are distance-dependent with pathloss modeled using the usual pathloss equation and shadow fading modeled as a log-normal distribution with zero mean and frequency-dependent standard deviation value. A combination of both is sometimes referred to as effective pathloss.
In terms of terminology, however, the combination of pathloss, shadow fading, antenna gain and minimum coupling loss is referred to as MACROSCALE FADING, distinguishing it from the time-varying SMALL-SCALE/MICROSACLE/FAST FADING.
I hope this clarifies the distinction between them. Just a try, the developers will provide further clarification if need be!
The Doctoral Dissertation in the link should also provide further guide.