To me that does not make sense, and I find it disappointing. Has anybody something to say about this?
It mostly makes sense. You did reduce the cooling flush, as opposed to increasing it.
Thermosiphons in an HX machine are not a very intuitive, nor linear, phenomenon. One very important facet of their behavior is that we care more about their impact when they're not siphoning, which is when we're pulling a shot. How fast the siphon was moving prior to the shot doesn't just affect the temperature of the group, but also the HX itself.
For a very fast siphon, the bottom (entry) of the HX would be at the same temp as the returning water, and the exit temperature would be below the boiler temperature, and the middle of the HX at some temperature in between. For a slow-moving siphon, the water at the middle and top (exit) of the HX would be near the boiler temperature.
So when you add a restrictor and do a cooling flush with a slow-moving (relatively) siphon, you're initially drawing more hot water than for the case without the restrictor. So you don't necessarily get a cooling flush that's proportionally shorter to the reduction in flow rate.
Thermosiphon Utopia is when the group at idle (while siphoning) is at the target temperature, and the flow rate of pulling a shot is identical to the thermosiphon flow rate.
-Ed