I don't think that frozen settled water at the bottom of the can affects anything, at least not for a while. In the type of can that Stevie uses the input hose discharges from the top. The PCV flow hits the bottom, then goes up again and gets out of the can through the output hose, also at the top. If the can is emptied before the real cold hits, it will take time to accumulate the amount of frozen water at the bottom to start spitting oil through the output hose. Plus, every time that the engine is started, the ice melts, and the water has a chance to escape as vapor.
Cold temperatures are the only reason the can catches water (brown milkshake-type stuff in the can). When it's warm, only oil of proper color is caught.
As for the purpose of the contraption, manufacturers would be glad to have an ability to separate oil mist from PCV gas flow. Their efforts are limited though by physical dimensions of the component that has to be part of the engine (to be maintenance- free). The condensed stuff has to go back into the sump. The user can't be forced to empty the can on regular basis...
I'm only familiar with 5.3l GM LS valve cover design with integrated separator, and quite a chunky stock PCV separator in 2.5 L Masda Skyactiv-G engine. Both are clear attempts at the problem, optimized for maintenance -free operation, but not highly effective. External catch cans installed after them catch quite a bit of stuff. This doesn't mean that external catch cans are more effective though, there's just no data. But they do help.
