Since the lean situation seems to be in the power circuit not idle, acceleration or circuit used for crusing(light load),I would study the passages taking the fuel from main jet and power valve to the main discharge nozzle.this circuit has a joint (probably gasketed) where the main well tube is attached to the upper body-a leak in venturi signal here will result in a lessened fuel delivery.
Then study the power valve and it's operating piston and spring-when accelerating lowers manifold vacuum signal the spring extends the rod,opening the power valve and introducing more gas AROUND the main jet into the main well tube--NOW that power jet passage to the main well tube has a calibrated restriction-it seems that could be a source of trouble causing leaning of the mixture when the power valve is needed to enrichen for acceleration.
The reason mixture is made richer or leaner when fuel level is changed is that the level in the bowl is the level in the main delivery circuit and when higher, then the same venturi vacuum signal lifts/delivers more fuel than at a lower level.
Just my thoughts from studying the Rochester b drawing and 50+ years involvement with carburator problems.
