Wide flat things are harder to tip over than narrow tall things. The "back corners" of a belly bridge are removed from the area where the stress would be greatest - at the back of the bridge.
Martin developed the belly bridge when the heavier gauge steel strings became popular, (and for dreadnoughts they still are for many people) to avoid warranty work I've been told or read (Longworth?).
If I was building for bluegrass I think I'd use a belly bridge.

I agree, that's why I think bridges don't lift from torsion stress though.

Most of the time when I see a bridge lifting, belly or rectangle, it's one side or the other.

If you think about the type of stress, IE bellying up, it makes sense that the ends lift first. Bridges don't lift due to the torsion stress or the strings pulling on them (for pinned bridges) , but rather the torsion stress causes the top to bulge in the center, and the BULGE is what causes the bridge to lift. One or the other sides pops up first typically, at least in all the repairs I've seen. I've even seen bridges where both wings have popped up and the center, with the belly, is still attached.

If I was building for bluegrass I think I'd use a belly bridge.

If you think about the type of stress, IE bellying up, it makes sense that the ends lift first. Bridges don't lift due to the torsion stress or the strings pulling on them (for pinned bridges) , but rather the torsion stress causes the top to bulge in the center, and the BULGE is what causes the bridge to lift.

It seems to me they belly no matter what.

A distinction without a difference. The bridge has torque on it from the strings. The greater the string tension and the greater the string height over the top, the greater the torque. The torque is the same...... whether the string is anchored underneath or tied on the bridge. The only difference is what happens when the back of the bridge pulls loose.

The bridge is a brace connecting the two X-braces, and a wider bridge is stiffer along its length. In other words, it is more resistant to bellying.
There is much precedent for using rectangular bridges on dreads with medium strings....just look at 1930's Gibsons. While most are short scale, there is the long scale AJ.
In other words, do whatever you like....just use good practice when gluing on a bridge. It is by far the most common glue joint to come apart.

The lifting/torquing would be greater

I've seen lifting bridges from bulges on the former, I've seen entire bridges ripped clean off, taking some of the top with them, on the latter as the strings wore through the bridge plate and top, and started to pull directly on the bottom of the bridge.

The bellying occurs mostly BEHIND the bridge,

For those of you using a rectangular shaped bridge what are the typical dimensions you choose?

Not true. Torque on the bridge is only a function of the string height and string tension.

Torque produces equal and opposite forces on each side of the bridge. The reason that the tops on most steel string guitars do not sink an equal amount above the bridge is because the X-bracing is stiffer there in relation to the bridge.