I guess I would design the circuit to theoretically work with half of the gaps bridged over. Then I could add or subtract some length to the open circuited stubs during tuning. If I were going to use wirebonds, I would model the wirebond as a short length or wire, or 200 ohm transmission line 5 mils long, or whatever your analysis software has available in its library.
If I were soldering over the gap, I would not model the gap, just extend the length of the open circuited stub to the the length of main stub, gap length, short pad length, and the fringing capacitance at the open circuited end.
Its not rocket science, you just farble around empirically until it works best. If it does not work at all, how the best set up of the stubs might indicate what is wrong and how to respin the board.
One other trick. Where the open circuited main stub meets the main thru microstrip line: You can wirebond across the valley, on the device side, or farthest away from the device, to "move" the location of the stub a little closer or further from the device.