At the end of that run it looks like there are mixed segments of walking and riding, with the telltale "buttonhook" pattern at many transition points. The question is: why switch back and forth? It's very remote, so doubtful that anyone would be out there.
Yeah, that one is a bit of a mystery to me too. I'm looking at the GPX in GoldenCheetah, to get a view of his stopped time (which Strava excises from its plots).
The first (and biggest) "buttonhook" is actually in the middle of a break. The RV stops at 8.3mi (I think we can be pretty confident that first 8mi section was all RV), the track is motionless for about 3 minutes, then there is about 5 minutes of wandering around the RV (forming the "buttonhook"), then 7 minutes again without motion. Finally he continues forward at walking pace.
The subsequent bursts are suspiciously the same speed as the initial 8mi section (and faster than his normal "running" pace), which suggests he hops back on the RV. But they're only 2 minutes long, and there is no obvious stopping at the fast/slow transition points, so it doesn't seem like he's getting into and out of the RV.
Maybe he's just hopping onto the ladder on the back for a stretch? Or maybe they're testing to see how realistic their RV-pacing is? Like "can Rob actually run *at all* at the pace we've been driving? Oh, shit, he can only do it for 2 minutes at a time? Eh, oh well, who cares, no one is ever going to scrutinize this data anyway!"