In a car radiator below the freezing temperature of water, the water part increasingly turns to slush. The glycol just prevents the mix from freezing hard enough to expand and crack engine blocks, etc.
But, ice is lighter than straight water and I haven't seen that mentioned or addressed here. Ice/slush rises in the system and then might be subsequently melted back into water the following day when the mixture is checked with a refractometer.
So. The glycol doesn't sink as was suspected, but ice/slush rises and then turns into water.
Similarly, separation is the greater of your problems because the slush will not pass through a small tube radiator and it can interfere with the thermostat and water pump bypass. If you fired up an engine for a short period of time, you can cause a log jam of slush in one side of a radiator tank. If the side tank has the radiator cap, I can see getting weak test results the following day when the slush melts.