We can view a hang glider (with attached pilot) either as a single rigid system-- in which case we must assume that the pilot is exerting whatever force is necessary to hold himself in a fixed position with his arms-- or as two independent bodies connected by a flexible strap and also by the pilot's arms. In the latter case, we don't constrain the pilot to be in a fixed position relative to the glider, but instead are interested in the force that the pilot must exert on the control bar with his arms to get the desired result, and the resulting torque exerted on the glider by the pilot. If we are interested in how the glider flies "at trim", then we want to know what happens when the pilot exerts zero force with his arms. In this case the hang strap can be considered to be the only connection between the glider and the pilot, which means the pilot's weight can be assumed to act as if it is located at the point where the flexible hang strap connects to the glider.
And where is the CG of the glider itself, in relation to the point of connection between the hang strap and the glider?
One way to think about this is to consider the case where the pilot is standing on flat ground with the wind blowing, with the glider producing enough lift to lift all of it's own weight and some of the pilot's weight, so that the hang strap is tight.
If the pilot exerts no "push" or "pull" on the control bar with his hands--if he lets go of the control bar entirely-- does the glider tend to trim to the same angle-of-attack when the wind is just barely blowing hard enough to keep the hang strap tight, so that the glider is barely lifting more than it's own weight, as when the wind is blowing so strong that the pilot has very little weight on his feet?
In other words, does scaling the wind force up and down change the glider's trim angle-of-attack?
Isn't scaling the wind force up and down exactly the same as scaling the glider's weight up and down while keeping the wind force the same, assuming that the glider isn't changing shape under load significantly? So if the trim angle-of-attack stays the same as the wind speed changes, doesn't this suggest that the hang strap, and the wing's net aerodynamic force vector, are both directly in line with the CG of the glider itself?
My experience is that, standing on flat ground with the hang strap tight as described above, scaling the wing force up or down doesn't significantly change the glider's trim angle-of-attack, suggesting that the hang strap and the wing's net aerodynamic force vector are both approximately in line with the CG of the glider. If anything, there may be some slight tendency for the glider to pitch up to a higher angle-of-attack when the wind is light, which may suggest that the glider's CG is slightly behind the point of connection with the hang strap.
Note that the hang strap and the net aerodynamic force vector from the glider are not exactly parallel to each other, as the pilot stands on flat ground in the wind with the hang strap tight. The hang strap's ratio of vertical : horizontal must be equal to (the upward lifting force exerted on the pilot) : (forward force exerted by pilot's feet on glider). The net aerodynamic force vector created by the glider has a ratio of vertical : horizontal equal to (weight of glider + upward lifting force exerted on pilot) : (drag of glider). Since the forward force exerted by the pilot's feet on the glider must be equal to the drag of the glider, it follows that net aerodynamic force created by the glider must be closer to vertical, than the line of the hang strap is. The weaker the wind and the less lifting force the glider applies to the pilot, the more the hang strap must be raked forward, while the direction of the net aerodynamic force created by the glider is fixed by the glider's L/D ratio so long as angle-of-attack remains constant.
But perhaps it is sufficient to observe that many hang gliders have been seen to fly rather well, at least in terms of basic pitch stability, when launched off a hill with no person attached. Usually this is not done intentionally, but rather as a result of the pilot forgetting to "hook in". This suggests that the hang point is quite near the CG of the glider.
As a practical matter, consider what would happen if the hang point were far from the CG of the glider. Imagine that you are running down a hill, connected to the glider by a strap. If the glider is only balanced nicely when the pilot's full weight is hanging from the strap, that means that when the glider is only lifting part of your weight, it is trying to pitch up, or to pitch down. This would be dangerous. But it the hang strap is near the CG of the glider, it will remain balanced to fly at a reasonable angle-of-attack regardless whether it is lifting your full weight yet, or not. The wing on a powered "trike" with wheels would seem to be somewhat free of this constraint, but it would still seem undesirable for the wing's trim angle-of-attack to change radically as the wing gets "loaded up" during the course of a take-off run, or for the wing's trim angle-of-attack to change radically for different weights of pilots, or as fuel is consumed during the course of a flight. This can only be avoided by putting the point of connection with the trike "body" close to the CG of the wing.
One more data point: at times I've added some rather heavy masses to the nose and tail of a hang glider. I always used the point of connection of the hang strap as the assumed CG of the glider for my weight and balance calculations, and the results were satisfactory with no observable change in trim angle-of-attack. Again this suggests that the hang point is quite near the CG of the glider.
Note that the hang point is adjustable forward and aft over a range of several inches, and some hang gliders even feature an in-flight adjustment of the hang point. Since a hang glider can be trimmed to fly in stable flight (in the pitch axis) with zero force exerted by the pilot's arms, at higher airspeeds (lower angles-of-attack) as well as at lower airspeeds (higher angles-of-attack) simply by moving the hang point forward or aft, it clearly is not essential for stable flight that the hang strap is located exactly at the CG of the glider.
It is common for hang gliders to tend to fly more nose-high as they get older, because the fabric stretches in a way that increases washout at the wingtips. Pilots compensate for this by moving the hang strap forward. Since the actual CG of the glider hasn't moved significantly, this is one more indication that it is not essential for stable flight that the hang strap is located exactly at the CG of the glider. Similarly, pilots who are heavier than the design target weight for any given glider usually find that the glider tends to fly at an excessively high angle-of-attack, again due to excess washout at the wingtips. Again, this can be compensated for by moving the hang strap forward.
At present I don't have immediate access to a set-up hang glider on a calm day so am unable to do a direct test by attempting to balance the glider at the hang point--
(Note-- in this answer when we talk about the point where the glider's net aerodynamic force vector is acting, we are using the (arguably outmoded) "Center of Pressure" model, rather than the model of an "Aerodynamic Center" plus a "Pitching Moment". The "Center of Pressure" moves as the angle-of-attack changes, and need not coincide with any physical point on the actual glider.)