Your setup
You mentioned the concentrated output from the power head has a more obvious effect on the tank than the diffuse intake on the sponges. This is explainable and perhaps best compared to a canister filter. You see, using the power head-and-pipes-concept as a manifold is similar to what happens when installing a canister filter with intake and outflow on either ends of the tank. You are creating movement
in the watercolumn, opposed to moving
the watercolumn. After all, you extract water from the watercolumn, put it through the pipes (alternatively, the canister filter) and then insert it back into the watercolumn through your output nozzle.
Here's the key to my point: this output nozzle has a limited surface area and physics state that moving a
high volume of water through a
limited surface area equals
high speed flow. Let us call this high speed flow 'jet flow': a narrow and fast flow of water.
We learnt that your output nozzle is inserting a jet flow into your watercolumn, which means the areas most affected by the jet flow are those directly adjacent to the jet. As stated above: you are creating movement
in the watercolumn, opposed to moving
the watercolumn. Those neighboring areas will become quite turbulent due to disruption of that part of the watercolumn, whereas areas farther from the jet remain relatively unaffected.
Illustrating my theory: the area behind the output should be relatively low flow.
Further illustrating my theory: after shutting off the power head, your watercolumn should be idle within a few seconds.
Comparison to my setup
Now, comparing my tank to yours to further illustrate my theory. There are two prime distinctions between them:
1. The surface area of the outflow. Your outflow is circular and perhaps one inch in diameter, whereas mine is rectangular (equal in shape to the watercolumn) and with a surface area equal to the surface area of the watercolumn directly adjacent to it. The same applies to the intake because that is equal in size to the outflow.
This means that any water from the outflow affects the entire watercolumn, because it
is the watercolumn moving through the manifold. In other words: I am moving
the watercolumn, opposed to creating movement
in the watercolumn;
2. My output affects the watercolumn from below whereas yours affects the watercolumn from the side. Your output is pointed toward your intake, not promoting any water other than what's directly in between them to move.
My setup eliminates turbulence because there is no movement
in the watercolumn and guarantees an equal distribution of flow through the tank (if you leave hardscape out of the equation, because hardscape of course obstructs the flow and alters the mechanics of water movement). I have in fact tested my powerhead on 100% capacity (75x tank volume circulation per hour) and the minnows handled it no problem. That is because the flow is diffused over the entire watercolumn.
Illustrating my theory: the water speed is roughly equal in any given part of the tank.
Further illustrating my theory: after shutting off the power head, my watercolumn takes a few minutes to settle (because
it was moving, opposed to creating movement
in it).
Conclusion (& solution)?
Do the fish mind? Your fish are happy, evidently. Breeding and all. All of the above was merely an attempt to give proof of concept. From your journal however, I could tell you are looking for solutions to your problems with different flow levels throughout your tank. Allow me to propose a solution to your problem with the flow, of course this is only theory from my part but it might be worth a shot.
1. Lower your output spraybar to roughly ground level, pointing it either straight up or slightly toward the right hand glass (away from the intake side);
2. Only place input holes facing the left hand side glass (away from the output side), so water is forced to the other side of the tank, affecting a maximum proportion of the total watercolumn.
Please consider the graphs below. The top one is your current setup (question marks indicating low flow spots) and the bottom one is my proposed solution.
