I'm suspicious that this piece has non-uniform density.
One bit of clarification is that the shift occurs over a range of about 20 RPM. It's pretty quick, but not bistable. Mid point on the shift is probably the RPM of maximum wobble.
But it wandered .0035" off center,
Centered ball within about .0001 to .0002" and intermediate wobble is gone.
long carbon fiber stem. Despite truing the stem, this top has a huge intermediate wobble around 420 RPM. Shortening the stem to 1.5" totally eliminated the intermediate wobble.
I thought for some time to all these your observations, but I am confused.
Normally, wobbling due to unbalance becomes more and more intense while the top slows down, and it reaches the peak of intensity when the top topples down.
This is the common situation, with my tops.
It doesn't matter the cause of the unbalance, whether the tip is off centered, or whether the distribution of weight is asymmetric.
There is not an intermediate peak wobble.
I didn't observe differences in this behaviour related to the lenght of the stem.
If something different happens, it is because there is a superposition of another movement, nutation or maybe precession, that makes things more complicated.
A few times I observed in my tops an intermediate peak wobble:
this peak wobbling was not due to unbalance, but to the top starting to nutate (nutation) spontaneously at a certain speed. The nutation, mixed with the unbalance, caused the temporary increase of intensity of the wobbling.
Later during the spin, nutation disappeared, as spontaneously as it started.
This spontaneous nutation is triggered by a large contact point, (it can be a weared spiked tip, or maybe a very large ball tip, or a ball tip with even minimal wear, because the contact point in ball tips enlarges very rapidly with wear).
Lack of lubricant makes the nutation to happen with stronger intensity and for a longer time.
I am not sure, but still I think that the intermediate peak wobble you see could be due to nutation:
it is difficult to say from a distance, but you can recognize nutation because it has a different speed from spin speed.
An easy way to check this, is with the tachometer. Measure spin speed as you do usually.
Then direct the tachometer at the upper part of the stem. Hold a piece of white paper behind the stem. The stem wobbling will intercept the light reflected from the white paper at intervals, and will let you know the speed of the wobbling.
If spin speed and wobbling speed are the same, this wobbling is due to unbalance.
If spin speed is different from wobbling speed, this wobbling is due to nutation.
If wobbling speed is twice the spin speed, and you pointed the tachometer at the center of the movements of the stem, this wobbling is unbalance, and the wobbling speed is twice because the tachometer reads two light interceptions for one round of the stem, one time when the stem goes to the right and another time when it comes back to the left. I point the tachometer a bit sideways, so I have one light interception for each round of the stem.
Precession is slow so it is easily recognizable. But at the very end of the spin, precession, if present, becomes fast, and could contribute to confuse ideas.
All the times I had an intermediate peak wobbling,
this was due to nutation.
I never had an intermediate peak wobbling due to unbalance alone If you are seeing something different, it is something I still don't understand.