I'm not a ballistics expert, but I'm not so sure that a bullet will increase in gyroscopic stability to be "stable enough" for terminal stability at long range.
I can't comment on "stable enough", that depends too much on bullet design and the dynamic stability implications of the specific design, which can't be good in case of the CEB, since they are awfuly long. Thus center of gravity and center of pressure have to be quite a bit apart, which makes it easy to overturn them (a general issue with extremely long bullets, not particularly the CEB designs).
On the increase I can comment though. Since velocity of the bullets forward motion is the main source of disruption of the bullet tip's ability to maintain a course parallel to the bullets flight path and S(g) is an indicator of the latter ability, a decrease in velocity has to result in an increase in S(g), if everything else stays the same, right?
The bullets rotation is the bullet tip's main factor in maintaining a parallel course to the bullet's flight path, it dampens out the pitching and yawing motions induced by air resistance due to the bullet's (minimally) asymetric form and imperfect center of gravity. If the rotational velocity decreases proportionally slower than the bullets forward motion, S(g) has still to increase, right?
The rotational velocity does decrease slower because the resistance it meets is just the surface friction of the surrounding air, the bullets body does not have to displace air to keep rotating. The bullet's tip otherwise has to displace air, usualy that much, that the air can't be compressed and pushed to the sides directly in front of the tip anymore, but is accelerated itself and has to be pushed in the direction of the bullet's motion.
In "Bullets - Complete Edition" by Beat P. Kneubuehl PhD, it is stated on page 104 that the S(g) of the 147gr Nato FMJ bullet is 1.37 at the muzzle and 2.29 at 300m. A 67% increase in 300m. I hope these numbers are enough to show the magnitude of S(g) increase in flight. If you compare the hypothetical S(g) of a Nosler Accubond fired by a short range hunter at a white tail 150m in distance out of some factory rifle, to the hypothetical S(g) a marginally stable (S(g)>1<1.4) longrange bullet has at 900m, I guess it's plain that the latter's S(g) at impact has to be way higher than the Accubond's S(g) at impact.
The Accubond (or any other stable (S(g)~1.4 at muzzle) hunting bullet) will usually stay stable at these ranges when penetrating the target. If the ability to stay stable after impact into and during penetration of a target medium depended mainly on the projectiles (S(g) at impact, the long range bullet would have to stay stable inside the target medium, too.
So what I wanted to convey all along is, that the rifle's twist doesn't have a great impact on a bullet's specific terminal ballistic at
long range since the bullet has already gone to sleep, as long the bullet is launched with an S(g) sufficient to keep it from keyholing at short range. At short range a S(g)>1.(4@the muzzle) is more important because you want the bullet to go to sleep as fast as possible to make sure it impacts point first without yawing and pitching alot, which could screw up the bullets terminal ballistic behaviour.
I hope this wish-wash is readable, makes sense and can give you a picture of my understanding of long range terminal ballistics. Please take into account that English isn't my native tounge.