Hello,
The variable time step was done in 2D for the finite volume algorithms and it ensures their stability with a clear theory. We would not know in 3D on what criterion or theory we would decide the value of the time step. The only theory we have is for the distributive advection schemes (N, PSI) and they organise their own time stepping within the user time step so that they remain stable.
With best regards,
Jean-Michel Hervouet