(Darn it, wildman beat me to it. Anyway, my long-winded version follows.)

From what I've observed:

When water freezes, it starts by forming a crust at the top with liquid water underneath. I assume this is because ice crystals are less dense, so they float to the top.

In a container, this ice crust forms an effective pressure seal once it reaches a certain thickness. When the rest of the water slowly freezes (and expands) it exerts equal force against the "ice crust" and the rest of the container wall.

For a container oriented vertically, the surface area of the ice crust will be smaller and thicker for a given volume of water. So, the container wall gets the maximum pressure possible -- and may burst.

For a container oriented at an angle (horizontally) the surface area of the ice is the largest possible, and so is thinner and weaker. It is likely more pliable for a longer period, so it may move and bow before it becomes an absolute pressure seal. So, there is less pressure on the container wall, and less chance of failure.

(I warned you, long-winded. Feel free to "flame" my ice theory. It's possible that I am hopelessly snowbound on this.)