I imagine the pivot (fulcrum) of the bar is close to its center so it is actually close to stable for any angle of inclination. On the other hand, in Iacopo's experiment the fulcrum is above the center of the beam, what would make it stable when the beam is horizontal like in a regular beam balance.
This I see as a part of the answer but still I think there is something else:
the fulcrum nearer to the center means a weaker force to keep the bar horizontal, but anyway it still should stay horizontal.
I think that if I build the same gyroscope as we see it and I test it, in my case the bar would stay horizontal.
It seems to me that there is some inaccuracy in the construction of Laithwaite's gyroscope:
Look at the very first seconds here:
A possibility is that the axis of the hanging gyro is not free to reach 90 degrees inclination from the bar; if the center of mass of the gyro with its axis can't stay exactly under the pivot with the bar, when he balances the bar, this will affect the balance. Even only 2 or 3 degrees of the gyro axis tilted outwards could be enough: this would have the same effect as the gyro was suspended to the bar, not where there is the pivot, but a bit more far from the center of the bar. It would be like there is a bit more weight in this arm of the bar, (the arm a bit longer).
This would be compensated with the counteweight in the other arm of the bar a bit more far from the center of the bar.
When the gyro spins, its weight will be on the pivot, which is closer to the center of the bar, so it will appear to be lighter.
I'm not sure that this is the explanation, but it seems possible to me.