Bookworm,
I don't have the knowledge to argue about how much water condenses out for a given temp drop. Suffice to say that I'm pretty sure that it doesn't matter what order you heat and cool, the amount of water in the air doesn't change (unless it rains which is another somewhat mysterious process). For a given pressure and RH at 5 deg you will get RH' at -5deg and RH'' at -25deg regardless of what route you followed to finally end up at the temprature.
I'm stepping into dangerous territory of guesswork from 1st principles here!!!: Your pointing out of supercooled droplets triggered a memory and got me looking up me old notes.
Something to bear in mind, is that when water condenses it releases latent heat into the atmosphere. The same is true when supercooled water droplets (SCWD) impact and then move to the stable ice state. This process of releasing latent heat (80 cal for deposition?) is the problem for ice formation, because if there is enough released heat to partially melt the now iced droplet, it will re-freeze and this it is the re-freezing that forms the dangerous clear ice. The portion of the SCWD that was instantly frozen is fluffy and aerated (rime ice) and probably gets blown down the tube almost instantly. At -25deg, significantly less of the SCWD is melted and re-frozen preventing serious ice formation. At -5deg the opposite is true and the majority of the SCWD ends up as clear ice.
Obviously, that last paragraph assumes that wing icing and carb icing are two sides of the same coin.
With that in mind, your point about how much time this takes is taken. But I guess if ice forms on the wing of a jet liner at 450kts TAS then 150kts inside the carb presents no real barrier. A quick and rough calculation of carb body at 150kts and airliner wing at 450kts (TAS of course) suggests that they both form ice in about 0.01 seconds!