If you look up Sagnac effect, you'll see it requires two signals to be sent in _opposite directions_ around a loop so they both return to the transmitter.
If you sit in Japan and transmit a signal along a single path to China and have it return to you, that's half of Sagnac. You also have to send the other signal in the opposite direction from the _same source_. Where can it go? Only along the same path.
There's nothing complicated here. This is simply not a situation that the Sagnac effect is supposed to occur in. Instead of scaling down the effect from the earth orbit, calculate it directly for this case and you'll see the answer is zero, not approximately zero, not 1, exactly zero:
time difference = 4 * loop area * angular velocity / c^2
time difference = 4 * 0 * anything / c^2
time difference = 0
Please identify which step you don't agree with.
I think the whole Sagnac business is a red herring. Your real point is that the signal travel time depends on its direction. Yes? Why not focus on this? MMX tells us the time difference should be zero, as you suggested. So the real inconsistency is between the predicted zero and the measured 90ns.
