Failure in nonlinear reports simulation success

[Martin Sjölund]

See the attached model, running with +postOptModules= (to disable tearing) means the simulation stops at time 1e-05 (the first step), but the simulation returns success. The same with tearing enabled.

Also, the chosen matching seems odd. It creates a non-linear system where a variable used inside both a sign() and sqrt call is being solved for rather than the left side (this is solved as expected when breaking the loop).

With +d=delayBreakLoop everything simulates fine. But the result is not the expected (the pressure does not build up).