For your position, the correct setting currently would be DST=NO, UTC +2h. Otherwise there will be a 15 degree discrepancy upon slewing in the position of the mount relative to the object.
There has to be a simple reason for that mount behavior. All I can say is I used my CEM25P again last night and it flips perfectly fine. My settings (in the hand controller) are to flip (not stop) if mount exceeds meridian by 4 degrees set in the hand controller. In EKOS I have set it to flip if HA>2 degrees (not hour angle). But I am also baffled by the Park position, those two problems may be connected.
Nico, why don't you take your telescope (and counterweight) off so the scope or mount won't get damaged by strange slewing and then figure out where your Park position actually is. Then set it to where you want it to be, purge, RESTART Ekos, set the new position, write and save it.
I remember reading a similar horror story on Cloudy nights once, but can't remember how that got solved. You may want to search there.
All I can say is that is a weird setting problem somewhere, it has nothing to do with the CEM25P itself or with EKOS. Both work perfectly fine together.
Can you also please post a detailed log.
Set the log to verbose. You probably won't have to select guider, fits, filter wheel, etc, but definitely INDI, Alignment and the mount. Then run a session in the simulator. You don't even need to pick a target, just slew close to the meridian (using the mount control panel in Ekos), then let the mount cross the meridian. To speed things up you can set HA to >0 in Ekos so the mount will flip as it crosses the meridian. You can work all this out using the simulator. No need to waste precious observing time.
Jo