steady-state fluids: femlin_e or femnl?
Some demos use rho > 0 and femnl for steady-state fluid problems, some use rho = 0 and femnl, and some use femlin_e.
E.g. demo_2df_axi_Poiseuille used rho > 0 & femnl, but these results differ from rho = 0 & femnl and femlin_e. rho = 0 and femlin_e appear to give the correct (analytical) solution for the outflow.
Which engine do you use, when and how? What is the added value for rho > 0 and/or femnl??
It further turns out that results with femnl/elup only match femlin_e for nincr = 1, and depend on nincr otherwise (see also issue #3)