"HPxRPM number ain't got nothin' to do with nothin"
True - HP × RPM is TQ × RPM × RPM, doesn't exist.
The "combination" is TQ × RPM, not TQ + RPM.
"The amount above peak HP to shift at is a function of how close the gear spread is."
And how flat the torque curve is, and how much change in resistance occurs with higher speed.
If torque falls off quickly after peak power (common in highly developed resonant systems tuned for peak power) shifting at or immediately after peak power produces better results.
If torque is high (but decaying) after the peak you can use a higher RPM figure than peak.
If torque below the peak is poor, even more RPM above peak is needed.
From a practical perspective, near maximum speed the engine must recover to an RPM level at which the power is almost as high, or the car will slow down ("the damned thing goes faster in 3rd than 4th"). Really fast cars, or peaky engines need a very close shift into high gear, and can use more "progession" (the spread in RPM drops through the gears, with the same 1st gear), which would have little value in a drag race.
An example of high progression, #1:
1st 2.50:1
2nd 1.55:1 recovers 62% from the 1-2 shift
3rd 1.10:1 recovers 71%
4th 1.00:1 recovers 91%
This is better than less progression with the same 1st and 4th, #2:
1st 2.50:1
2nd 1.65:1 recovers 66% from the 1-2 shift
3rd 1.25:1 recovers 76%
4th 1.00:1 recovers 80%
#1 has less total power under the RPM curve than #2, but is still has higher top speed, because the critical 3-4 shift is the only one that occurs when the car is aero-limited rather than traction-limited.
Why not just use all close ratios, like #3?
1st 2.00:1
2nd 1.40:1 recovers 70% from the 1-2 shift
3rd 1.10:1 recovers 79%
4th 1.00:1 recovers 91%
For some transmissions it doesn't exist, or $$$. For some, the much lower numerical 1st gear (i.e., 2.00:1 instead of 2.5:1) makes launching the car very difficult (clutch slip, converter heat problem, plug fouling, spark curve), and the engine must develop more power at low speed to overcome this.