If you connect a floating applied part (the bit that connects to the patient) to earth the applied part is still isolated via the instrument - the test fails because you're applying earth-referenced mains direct to to earth, not testing the isolation of the applied part.
In practice leakage would not flow through the patient via the instrument connected to that applied part. The fault-current or "leakage" is actually bypassing the applied part connected to the patient and flowing to earth - you've made an external circuit.
Otherwise you must be talking about earthing of applied parts inside the DUT - which in that case they wouldn't meet the BF or CF requirements in the first place. Where do you mean connected to earth - internally or externally?
If a device using an external isolation or seperating transformer, for that matter, is used and the applied part is earthed then tested with earth-referenced mains on the applied part test, you still end up with the same result - a fail and excessive current flowing through the connection to earth.
I don't think it was a very good example to use, that's all, irrespective of the degree of electrical isolation being discussed.
