I've seen this happen before. If the inductor doesn't have a "device description" in the rule file then there may not be any special layer over the inductor that breaks the connectivity between the pins. The concept is very similar to resistor devices in LVS. Especially similar to metal resistors in LVS where a "resistor layer" is drawn over a section of metal and it is subsequently recognized as an LVS device where the resistor seed body is formed where the special device layer crosses the metal, and the metal in that region is also cut so it will no longer be considered conductive.
The special device layer for the inductor is what normally cuts the metal so the pins won't be shorted.
LVS BOX purposely doesn't remove metal, it just removes devices. EXCLUDE CELL removes everything (devices and metal) but that's usually too extreme because the whole cell is gone and nothing is left in it's place.
I'm not aware of any fix besides using a special layer over the inductor metal so the pins won't be shorted. If you don't want to modify the rule file but you are able to modify the layout and you are willing to use LVS BOX then maybe you could check your rule file for the existence of metal resistor devices... if there are metal resistor DEVICE statements in the rule file for the layer of metal being used in the inductor then you could draw the metal resistor identification layer over a portion of the inductor and then at least the metal pins would be separated for the purpose of connectivity recognition even though the wrong kind of device would be found in the cell (a resistor instead of an inductor) but then the LVS BOX would causes that resistor to be ignored during LVS comparison.