Excellent question.
Here's my extremely imperfect understanding. All corrections more than welcome (Jonh, Suitbert, you know who you are
).
The "center of gravity" displaces in a normal balance even when the balance is horizontal, and I'm assuming this is what your graph shows. This is due to the attachment points adding a variable force into the spring as it oscillates. It creates a problem as the balance staff is forced against the pivot in a variable way, whcih introduces errors. The double balance spring corrects this, if I've understood correctly. The variable forces due to the points of attachment also mean that the frequency isn't constant at different amplitudes. I'm not sure if the double balance compensates for this: intuitively I don't see how it would, but this is just guessing.
When the balance is in a fixed vertical position another set of forces come into play due to gravity. The balance staff has a permanent force pulling it against the pivots in addition to the forces due to the points of attachement. The balance spring has a permanent force that makes it "sag" and that has to be added to the variable forces. This makes the "center of gravity" behave differently. It's not clear to me how the double balance spring compensates for this, as both springs are "sagging" together in the same direction and I don't see where a compensating force comes from. I've read the Moser blurb several times, and I think they're being slightly disingenuous in their comparison with the the tourbillon.
I'm going to guess that the double balance spring removes the errors inherent in the spring and so makes the watch much easier to adjust across positions.
Like I say, educated guesswork.
As for which is better, as no manufacturer will ever release data we'll probably never know.
nick