Gizmo rewrite and bug fixes

Rewrote Gizmo so that it no longer needs the particle time step during the flux computation. This breaks manifest conservation of mass, momentum and energy, but makes the scheme more similar to the other hydro schemes. Got rid of a lot of unused particle variables. Fixed a bug in the way the velocity for inactive particles was set in the flux computation, and a bug in the way the interface area was calculated. The square test, which was broken, now successfully runs again.