I can't see that it makes any difference whether the N-E leakage is 'unavoidable' (due to connected equipment), in a 'sound circuit') or due to an N-E fault'. In either case, with the 'usual' (in my experience0 test button arrangement, that leakage will add to the residual current, so that , with 9 mA leakage, the device should trip with a test current of 21 mA or more. With the 'other' arrangement (per diagram you posted), what you say is right - in that case, with a 30 mA test current and 9 mA leakage (of any cause),the device might not trip.In a sound circuit if the test current is 30mA then if, say, 9mA is returning via this leakage then the RCD will not trip even though the 9mA may be unavoidable acceptable leakage.
BTW, I think that the test current is usually about 2.5 times the IΔn - hence about 75 mA for a "30 mA" device.