A tip is to have a look in the model run data file which tracks the modelling performance, it usually gives a good idea about where the problem lies. It automatically saves a text file in the same location as your MIKE3 setup file.
But, as for your problem; The initial WL is above the downstream boundary condition I believe? Just try to set the initial value as a constant surface elevation instead, which correspoonds to your downstream BC. Also, if you use a water depth, you get a water column up on your channel banks as well when you use a trapezoidal channel, which might cause instabillity. I would try that first (as your model is very small it will fill up right away anyhow).
Best regards,
Jonas