I checked with the team on this, and what we *think* is going on is this: the standard for "black" in video is actually a very, very dark gray. i.e., there's still some luma value in the signal -- it's not zero. A true zero luma is called "blacker-than-black".
In the case of this content, I believe the video file itself is in a 2.40:1 aspect ratio, while the movie is actually 2.76:1. (This is a very unusual ratio; only a few films use it because it's super wide. Examples include The Hateful Eight and Ben-Hur.) Anyway, because the content is in that higher aspect ratio, there are black letterbox bars encoded in the video frame. The area outside the active video image is rendered with a luma of zero (blacker-than-black).
So, what we're seeing here is the blacker-than-black bars at the top and bottom meeting up with the black letterboxing that's part of the encoded video. Now, when a display is calibrated, black and blacker-than-black should appear the same. Otherwise, you have what are referred to as "elevated black levels". So it's likely that a display calibration would resolve this issue.
I've also inquired with the team whether we could have just encoded this file in a 2.76:1 ratio, and whether that would have resulted in the entire unused area being rendered with a zero luma value. That might be desirable, but I'm not sure if the media engine has assumptions built in about what content ratios our files are encoded in.
Hope that helps to explain.