With respect to a PV inverter, you've still got to consider the fact that the device that is closing the circuit is a device that only closes the circuit if all the voltage and frequency parameters are checking out every handful of cycles. So yes, the current will subtract, but that physically entails the voltage going out of whack, and pretty much as soon as that happens the circuit will be opened and the fault cleared. And the GFDI and anti-islanding shutdown requirements for the devices are really fast, faster than max trip times for thermal magnetic breakers, as far as I can tell.
If you have a fault that subtracts less than the inverter output current, it will get caught by the GFDI; if the fault occured while the inverter wasn't operating (at night), then the inverter won't even turn on and there will never be an AC fault. If you have an AC fault that is larger than the inverter is rated for, it would pretty much have to be from some freak occurrence causing sudden major damage to the wiring while it was operating. (Say, a tree falling on the array). A blown inverter is likely to be the least concern.
Finally: Even if Electric-Light is completely right about the possiblity of a non-isolated inverter getting destroyed by utility AC flowing through it, that is a safer, less costly outcome than an isolated PV system setting a building on fire because the first fault in the grounded conductor was never detected. The latter has been documented publicly. The former, to my knowledge, has not.