I have wondered about this at times as there are two ways the rotation of the bullet is considered. Mentioned is the rotations per second, the other is the actual path of the bullet. with rps the spin would actually become shorter as the bullet slowed, since the resistance being met by the bullet affects the forward motion of the bullet with very little affect on its angular rotation. If the bullet follows a path it would indicate that the rate of spin per distance would be constant and as the bullets forward velocity slowed, so does its spin.
It would seem that the rps would be the "right" answer, but if so, then bullets with marginal rates of twist for a particular bullet would become more stable as the velocity decreased. But this is not the case. If a long bullet is fired in a marginally twist barrel the result generally is that it will be stable for a few hundred yards, but unstable at longer range when it has slowed. So I don't know what the correct answer is, which is why I continue to wonder about it, but in the process, I follow the formula mentioned above and follow manufacturers recommendations on the optimum rate of twist for their bullets.