To check the whole situation a bit closer I now measured the slack in the linking rod.
To do that, I mounted a dial gauge with the tip at the lever of the throttle shaft (see photo) and set it to zero. Then I pushed slightly against the lever. The slack was clearly noticable. The gauge showed 20 hundrets of a millimeter = 0.2mm.
Now please compare: The adjustment screws have a 0.6 mm thread. That means that one turn makes up 0.6mm. That means that the slack in the linking, if coming into play, de-syncs your carbs as much as 1/3 turn of the sync screw would do!
We all know that a third turn of the sync screw is very very noticable and lets the engine run very uneven.
I think the only reason why that issue is not so noticable is the fact that this effect (which is caused by the accelerator pump) starts only when throttle is a little bit open and the de-sync effect ist no more that much noticable.
: Thank you for the hint, but the bushings are there. Maybe if they were new, the effect would be less, but the problem remains (if the right spring is not turned).
: Why would Honda not mount the neutral switch at a place where we could change it w/o dismounting the whole engine? Why does Honda build the neutral switch in a way that the lamp sometimes glows permanently? Why would Honda not mount the central stand in a way that it does not bend? And so on... I think even Honda does not construct every part of every bike perfectly... And we have sometimes the chance to improve things ;-) Like here at the carbs...