1. I know what an accelerometer measures. Force. I know physics.
No, it doesn't measure force. It reports the instantaneous rate of change in velocity at any given point in time. If you expect those values to give you any information about speed, position, or orientation, you need to integrate the data over time.
If the structure is moving the accelerometer around, flexing of the structure is going to result in less than accurate measurements, due to vibration.
If that is what you are seeing, you need to make the structure more rigid.
2. Yes, the noise is caused by motors.
Electrically, as noise, or mechanically, due to vibration caused by the movement of the motor shaft not being perfectly concentric or because of less than perfect bearings?
4. I tried already using rubber mounting on the accelerometer. I tried also the PU gel. (I got improvements, but not as much as I want)
5. Do you know a better solution than all of these? I was advised some time ago to sink the board into some form of fluid which would reduce most of the vibrations. What do you think/know about it?
Perhaps I've lost the connection with what you are trying to measure with the accelerometers. I suspect that it is the motors that need to be de-coupled, mechanically, from the structure, with rubber or spring mounting. Isolating the accelerometer from the structure is like having sex wearing a condom.
On the other hand, I don't really understand what you are trying to measure or what you are trying to use the data for. So, maybe the condom is appropriate. I don't know.