Having worked as a chemist for a few years and having used liquid nitrogen(LN)quite extensivly, there are two problems with the LN. The first one was previously mentioned, and that it displaces room air very quickly when it evaporates. Thus creating a hypoxia (no oxygen) problem. However, with this application, you probably wouldn't use enough to cause a big problem (although still something to consider). The second problem is the more serious one. If you are using any kind of open system to room air, then your problem is that LN is cold enough to cause condensation of oxygen to Liquid Oxygen (LO). LO is used in the space shuttle as an oxidizer for the fuel (there is no oxygen in space and it has to come from somewhere). LO is extremely explosive and when it goes from liquid to gas, it expands much more violently than does nitrogen doing the same. In fact it is an explosive expansion (I have witnessed this personally being across the room when this happened to someone else). Now, couple the explosive expansion with a hot filament from a MH lamp that bursts when the oxygen explodes, now we have fire. A very hot and fast burning fire with lots of oxygen and fuel (your house)! Now you have no tank, no fish, no corals and no family because everyone has been burned to a crisp. I would scrap the LN idea (although its nice to see someone brain-storming new ideas).
The dry ice idea could be done by using a slurry of dry ice and isopropyl alcohol. This slurry will cool to (-78) C. But again is the fire hazard with the alcohol, not to mention you have to have specialized containers to hold the mixture to prevent condensation and these are very expensive.
I would stick with the chiller or milk jug method for now. But keep thinking, thats how new things are invented.
Jay