Okay...... after sleeping on it, I think I can better explain how you can have an inertia chassis dyno sheet where the power curve is not at all representative of what the engines power curve would look like, including have a big spike much farther up than where the engine tq peak is.
This is basically all pertaining to running a high stall converter on an inertia dyno.
One of problems is how the graph is displayed.
Without any time element to the graph, it doesn’t paint an accurate picture of how the run progresses from start to finish.

First we have to understand that the torque converter is the coupler that connects the engine to the load....... and that the stall speed determines at what point(rpm) full load will be applied.

Secondly, on an inertia dyno, the power is determined by the rate of acceleration of the drum.
Nothing is “measuring” the power output.
Accelerate the drum at a higher rate, power displayed goes up.

If you did a test with these parameters, you would get a dyno sheet that showed very low engine output at the lower rpms of the graph(and have no bearing on what the engine power would be at those points).

Assume the engine in the car has already been on an engine dyno and makes peak tq between 3000-3200 rpm(not that this matters, I’m just using it as an example to point out that even with a peak tq of a known lower rpm point, a loose converter will show peaks at a higher point than where they actually happen).
The car it’s installed in has a converter with an actual flash stall speed of 5000rpm.
The test will be conducted from 1200-5500 rpm.
The car will be in high gear, throttle held steady at 1200, drum speed stabilized, you push the button to start recording just before you push the throttle down....... the motor accelerates to 5500rpm, you push the stop button.
The data recorded will be from 1200-5500rpm.
On the tq/rpm graph, the point where the full load is applied to the motor(5000rpm) will be most of the way to the right side of the graph.
And there will be a big spike in output right at about that point....... because that’s where the converter started to couple, and the big change in acceleration to the drums occurred.

You could pick points along the graph at the lower rpms where the numbers were much much lower.
That’s because with the converter slipping so much, there isn’t much change in acceleration rate at the drum. The converter is still flashing up to stall speed and is not very effectively coupling the motor to the load.
The graph would have a lazy slope from 1200 to the point near where the converters stall speed is........then as you approach the stall speed, converter couples, drum rate surges, and you get a spike on the graph nowhere near where the engines actual tq peak is(in this example).

Here’s why this is not an accurate picture of how the run really happened, and why looking at a graph of the run with a time element attached helps tell the story.
When the start button got pushed and the throttle went to the floor, the motor immediately went right to the 5000rpm stall speed.
If the motors tune was sharp, that would happen in less than one second(1200-5000)....... yet it takes up 3/4 of the space on the tq/rpm graph.
If you plotted rpm on the y axis and time on the x axis, the line would go almost straight up from 1200-5000(and would all happen way at the left side of the graph), then at whatever rate the power being produced from 5000-5500 could accelerate the drums.
If the motor was making pretty good power, this whole run from start to finish would be maybe 2-2.5seconds...... because the load would have only been applied from 5000-5500rpm.

It’s not a long steady pull like the tq/rpm graph depicts, and the numbers below stall speed aren’t representative of engine output whatsoever.

The OP’s motor should have an engine output torque curve similar to the one I posted the sheet for.
That one had peak tq at 4000 rpm. At 5000 it’s already down 93ft/lbs from peak.
If you chassis dynoed that motor in a car with a 5000 stall speed converter, that’s where the highest torque will be displayed, since the torque curve is already well past peak, and that’s the lowest rpm where full load can be applied.

The other thing the rpm/time graph does is give you a nice picture of what the flash stall is.

What I’d like to know from the OP is, what’s the flash stall speed of his converter.
If you haven’t tested it yet, try this......
Disconnect the kick down linkage, put the car in drive...... let it up shift into 3rd...... hold the throttle at about 12-1500rpm and let the speed stabilize...... then floor it, and observe how high the tach needle “flashes” to.
For all practical purposes, that’s your flash stall speed....... i.e. the lowest rpm that full power can be applied to the wheels.