I've also reproduced this issue in 2022.3.26f1, I use `SmoothDamp` in various places in the game and there is also pause functionality that sets `Time.timeScale` to 0.

I've never before noticed NaN to occur as a result of `SmoothDamp` but today I paused/unpaused the game in rapid succession and suddenly found various unrelated `SmoothDamp` calls in the game causing NaN propagation at unpredictable times (unpredicable but it very consistently does so if you pause and unpause in rapid succession for a few seconds).

This related ticket seems to have been set to Won't Fix but I don't think that's acceptable: https://issuetracker.unity3d.com/issues/mathf-dot-smoothdampangle-sets-currentvelocity-to-nan-when-time-dot-timescale-is-set-to-0-and-target-is-close-in-value-to-current

Every game that uses `SmoothDamp` and also has a pause feature (which is probably every game that isn't multiplayer-only) runs the risk of this issue happening.

If this is indeed expected behaviour and you want users to be more careful when using `SmoothDamp`, that is fine, but then the `SmoothDamp` documentation needs to be updated to say so. As a user you can reasonably expect `SmoothDamp` to just return the input value if `Time.timeScale` is 0, like it seems to do correctly *most* of the time.

As a user, I would expect `SmoothDamp` to internally check for invalid intervals and just not modify the input value in that case. Quaternion similarly checks for invalid conditions and even logs to the console if you use it incorrectly, `SmoothDamp` is such a core feature of the engine that I would argue that it deserves the same level of caution.