Have noticed a similar issue after applying the HUT Ultra Gloss white plastic polish to a CA finish. Use an air gun to blow out any little voids with sawdust residue after sanding to 600 grit and then again if the white polish appears in the voids after final polishing. My issue is why does the CA finish, 5 coats thin, sand @ 600 grit, 3 coats medium then wet sand to 12000 still leave those minute voids in the blank? Thought the CA finish would fill those voids.
Remember that if the lathe is spinning, physics is working against you filling any void--you're trying to get a substance
into a void, while the forces at work are trying to expel anything they can.
Additionally, the thin CA is so thin that it either soaks right through or immediately gets thrown back out. It may seal the edges of the void, but it's not likely to fill the void, even with multiple coats. And the medium is too viscous to make it into the smaller voids with the lathe spinning (and often even with it off!).
With all that in mind, if there's anything I don't want coming through to the final surface, I try to fill it with the lathe off. This takes more time and a lot more patience. I usually do a "seal" coat or two with the thin CA, then a single "fill" coat with the medium. In my experience, if I try doing multiple coats of the medium, I increase the chance of contamination, or simply of getting a bad cure--where the ambient moisture is trapped between coats and then cause the very white spots we're trying to avoid. But since I'm using more glue to fill the void, I have to allow for more time for curing, even with accelerator spray. Nothing worse than thinking a void is filled and the glue is cured and then getting sprayed with CA when you turn/sand past the thin layer of cured glue and expose a pool of uncured glue.