Now, we dump the frame if it has been already transfered - this is because if we pause waiting for the memory to get free... we create a who for the output (tx) which is bad for PCS....
If the frame has not been yet transfered, we can wait for RTU decision

Not sure this is the best solution

ALso, I added handling of the almost-impossible situation when we get the nonmem when we are in the cycle to decide whether to transfer or not... nusty