The reason I point out standard usage and the design parameters of an incident light meter is so that people will have a better understanding of the tool and how it is designed to work. Once that is understood, one understands better what's happening when departing from the method for which the tool is designed (i.e. pointed with incident dome toward camera in the same light as the subject). If a non-standard method of use is portrayed as standard usage, it can prevent that understanding of the tool's design and prove confusing, especially to beginners. Of course everyone is free to use their tools as they see fit. Understanding the design will help with making intelligent departures. See Phil Davis and his incident metering method in Beyond the Zone System for one example, where he shows how and why one can determine subject brightness range with an incident meter in many/most circumstances.

In practice, under many circumstances there may be less difference between standard readings pointing an incident dome toward the camera and away from it than many people assume, and film latitude (especially negative film) often easily covers many of those departures. The domed incident meter does a great job of integrating a reading in varying lighting circumstances as Dunn and Wakefield (and many others) have mentioned. It's more comprehensive a tool in standard usage than many people give it credit for. The best thing is for people to point their domed incident meters on and off the lens axis under varying lighting conditions and pay attention what happens to the reading.

The photo linked to by the OP was made with an FA, likely with early matrix metering. If the appearance on screen is indicative of what's on the negative, then I'd say it was underexposed and overdeveloped. The skin goes from blocked up highlights to blocked up shadow in the space of a few inches. To me it looks very much like film that was 'pushed' at least a couple of stops.