Sorry for the small delay @bertrand!
Some things I would recommend to you:
1. Could you try using the standard wall boundary condition and not the customizable one?
2. The simulation has to run for a longer period of time - only 20 timesteps are performed in your simulation. An incompressible flow case requires something more than approx. 200 to get to steady-state. An ideal number will be 1000 to 2000.
See if these changes have a significant impact on your simulation and let me know how things go!