The whole argument is about how the change in heat does no depend on frames. Heat is a quantity defined for the system of particles.

Individual particles' kinetic energy might be constant after averaging over short time scales, as in the specific example you gave, but this is not required. The conclusion has been drawn based on much more general reasoning.