You are theoretically right, but the load transmitted will be proportinal to the deflection, not necessarily half the load on the member. Framing members are designed for a full live load plus a dead load, and the full live load is rarely achieved in practice, although it would be more so on a roof in a snow area. In actual practice, in the pictured example, if the rafters were designed to span from ridge to wall, a very small load might be transmitted to the truss, under full live load conditions, but it is probably not enough to worry about. If the “designer” (if there was one) was counting on those struts to transfer load to the trusses, then yes, there is a big problem.
In a house where there are many partitions but only one load-bearing partition, the same possibility exists over all the non-load-bearing partitions.
Theoretically, second floor or ceiling joists would deflect and place a load of some sort on the non-load-bearing partitions, but do we design for that? We would spend more time calculating than drawing the plans if we did, and that is only if it were possible, and then we’d need all sorts of supports in the basement, or some mighty heavy first floor joists, to deal with all those loads.