With our combi DHW still playing up, I finally pulled the flow sensor apart.
It's a plastic impeller with a ring magnet set into a groove. The one I extracted had an incomplete (I am guessing broken from looking at the photos on eBay ?) part circle of plastic around the magnet.
So the idea is as the impeller spins so does the magnet and some sort of hall-effect goes on in the housing that is translated into pulses for the control board. (Based on the YouTube videos where a helpful chap insists you need a "Hertz" multimeter ...).
All of which seems an ingenious and over complicated way to solve the problem with as few moving parts as possible.
No idea WTF the magnet wasn't embedded into the impeller. I guess they have to sell spares.
But I keep thinking there must be an easier way to determine flow ? Surely if you have a venturi, or series of venturi in a housing of known dimensions, then you could determine flow from pitch of water passing through. Even if you had to shape a baffle to give it a hand to whoosh ? Combine that with temperature and you'd have a pretty decent sensor that would be more reliable, surely ?
Returning to the failed sensor in hand I wonder where that tiny fragment of plastic ended up ? It wasn't anywhere in the housing.