I've heard this question before... I will put forth my theory...
If the total of the thrust from the fly's wings reacts exclusively on the bottom of the bottle then the scales will likely oscillate heavier and then lighter. This would occur because the fly’s wing action does not produce constant thrust.
Though, I think on average the scales would read less because not all of the wing thrust would act on the bottom of the container. Some of the thrust be dissipated in the friction of the air fluid and some would act at an angle other than perpendicular to scale. The total of the flight thrust will not be applied to the bottom of the container.
In conclusion, I hope that you folks are not spending too much of your employer’s time on this…