Not to mention, accurately measuring independent download times on a web page (or a native app screen) with 40 assets is next to impossible because of the overl

Not to mention, accurately measuring independent download times on a web page (or a native app screen) with 40 assets is next to impossible because of the overlapping request, persistent connections and all sorts of other confounding factors. The actual latency you should measure for such a download is the total time taken to get to a usable view for the user. How to do that is left as an exercise for the reader ;-)