Several things to consider here.
- Starting from zero fluid flow, with a non-ramped inlet condition will cause an initial pressure wave going from the inlet into the domain. Consider either starting with a ramped inlet condition, or an initial condition that is non-zero (for instance generated with a Stokes solver).
- Using strong pressure conditions is not something I would advice, see for instance: https://asmedigitalcollection.asme.org/appliedmechanicsreviews/article-abstract/59/3/107/465438/On-Boundary-Conditions-for-Incompressible-Navier
- Have you tried with a direct solver, rather than gmres and hypre?
If you have a similar setup in legacy FEniCS, could you provide that code as well?