Why is it not a issue (neutrals and grounds connected together) at the service disconnect but and issue at the sub-panel? (Potential to carry neutral current)
It's really just the convention in this country and much of the world, but essentially under the NEC jurisdiction they are always kept separate in the circuit wiring, while on the utility network one conductor (the grounded conductor) serves both purposes. The service is where they are bonded together because that's where the two networks meet so that's where the EGC splits off from the grounded conductor under the NEC. (They are also bonded together at the source of a separately derived system.)
The intention under the NEC is that no current flows on EGCs and other exposed metal within buildings, under normal conditions. The EGC has to be connected to the grounded conductor at one point in order to serve its purpose, but it should only be connected at one point to avoid the EGC becoming a parallel path for current under normal operation.
The system we use is described on this
Wikipedia page on earthing systems, it is the TN-C-S system. The protective devices and rules we use (breakers, GFCIs, EGC sizes, etc.) are designed around this system. In countries with different earthing systems they need different kinds of protective devices.