Consider that the drivetrain is already a collection of several springs with only a couple of dampers: the friction of the clutch and the friction of the tires. The springs are primarily the three long driveshafts and the tires.
Also keep in mind gear ratios. For example a ~1 rev duration engine torque fluctuation is experienced through only 30 - 90 degrees of rotation by the half-shaft, wheel, and tyre. Assuming a gear ratio spread of 12:1 - 4:1.
On the engine side, the crankshaft loses something like 12% of its angular velocity with each upshift. This braking arrives mostly from the ratio change itself. The energy from that deceleration is necessarily absorbed mainly by the driveshafts and tires. Conjecture: when this shock absorbing system isn't damped well enough it can be heard in recordings as low frequency shudder. Conjecture: this undamped spring oscillation sometimes makes its way through the tire and a similar low frequency pattern can be seen as dotted tire marks on the track surface.
Regarding seamless shift duration: I see mention of times in the .005 - .010 s range, or sub 1 engine rev. I don't know what this is a reference to. The actual figure should be closer to zero. There is no break in torque delivery. I'm fact, it should increase during the upshift as the engine slows 12% within some milliseconds as the SST dogs step, or slam, into the next gear.
I think the milliseconds that people are refering to when they speak of a shift, or a gearshift, in this context, may actually be the *effect* of a gearshift. The bending of the springs, as it were. The gearchange proper should be instantaneous with these seamless mechanisms. The bending and deflection of all the parts connected to them is what lasts a few, perhaps many, milliseconds.
That exact duration is something I'm unaware of. I suspect there can be a lot of variance. It does not need to occur within one engine revolution, but it might. It might also be distributed across several revolutions, since gearing combined with drivetrain elasticity permits this.
Regarding clutch slip: is the hydraulic clutch actuator expected to operate at these timescales? If it is too slow to directly interact with the gearchange, it may be acting more slowly across the aftershocks and harmonics of the gearchange; effectively as the damper for the spring that is the driveshafts and tires i.e. acting as a friction damper that can transmit full engine torque while breaking free to truncate peak torque fluctuation resulting from rapid engine speed changes. Therein may be a point worth clarifying. Pressure upon the clutch plates need only be reduced to absorb torque spikes while still providing enough pressure to drive the gearbox slip-free. Perhaps that is the essence of the Honda paper.