This is rather odd, considering how manual malware reverse engineering works. Usually when you get a sample to dissect, you already know that it's a bogey. So it not doing what it's supposed to do is a quick way to become even more interesting, and finding that reason shouldn't take a good AV researcher more than an hour, tops.
It also doesn't really add to the complexity of the analysis, creating/copying a handful of documents into your VM isn't that big a deal, what you'll probably do is to clean up, copy the files in, create a new base image and run from there. The delay this would cause is minimal.
This as a hurdle for human researchers makes rather little sense, to be honest.