In order to obtain d'Alembert's principle, one must exclude situations in which constraint forces do virtual work. Actually, not individual constraint forces, but (according to the notes mentioned in the title) it is sufficient for the total sum of virtual work done by constraint forces to be $0$. The professor mentions 3 cases, from the least to the most general. The least general being the case of a single constraint force acting on a single particle doing no virtual work (aka normal force), then the case of a single constraint force acting on several particles (for example the tension of a string acting on two particles) and finally the most general case of several constraint forces acting on several particles.
There is no example given of this final case and the author explicitly states that he "admittedly cannot think of an example". I admittedly cannot either right now, and I need to continue with the course, but I am curious enough to ask.
To be as clear as I can: I'm asking for a case of a set of constraints, which even though the forces of a single one of those constraints does net virtual work on the system of particles as a whole, that work gets cancelled out by the work done by some other forces corresponding to some other of the constraints.
I guess the system should be a mechanical one in principle, but other sorts of systems are welcome as well.