rigpiggy, I'm aware of the bypass valve in the fuel side that maintains constant fuel pressure differential across the metering valve but has no air connection, and the minimum flow stop on the metering valve itself, but you seem to be talking about a separate Py-operated minimum flow valve. Am I understanding you right? I would expect though, if there were a second minimum flow valve, there would be a second minimum flow adjustment somewhere.
I found a different explanation
here. The cross arm that connects the bellows to the metering valve runs inside a torque tube that is connected to the cross arm on one end and to the casing on the other. That torque tube is the (torsional) spring that I couldn't find earlier, and is rigged to bias the metering valve to the low flow stop.
So, whenever Px = Py the bellows does indeed tend to collapse and increase fuel flow, but at low absolute pressures that tendency is overcome by the torque tube contribution. When those pressures get close to the normal operating range though the torque tube effect becomes negligible and the bellows respond to the Px/Py pressure ratio.