Have 1 or many.
I have (will have) a potentially large group of static triangle collision actors, terrain, mountains, etc that needed to be created/cooked at startup. Currently Im combining all their triangles and indices into two large lists, transforming vertices (at startup), cooking and creating a single actor to represent the world. There is some obvious load time to do this.
In terms of startup/load time, Im wondering if there would be any benefit to creating a static actor for each object, passing a transform with each objects triangle geometry, and cooking multiple static actors (rather then combining into 1 actor).