As far as I'm aware, that is how the software is set up to be. Would definitely be more efficient utilising all cores, but unfortunately there doesn't seem to be a work around for this issue... That I'm aware of. Will definitely keep looking, and will post something up if I find a way around this. Quite interested as I'm running electrically large simulations too.
In the meantime though, if ridiculously long run-time is a problem for you, you could try using other solvers (saw your screenshot using Time domain solver) to speed up those simulations. Different solvers cater to different problems and T-solver can take quite a long time due to the fact that it utilises volumetric meshing as compared to surface meshing like the F and I solver.