Yorg: I think a feature could be just as safe as purpose models if it’s designed right. While the purpose model workflow is a good workaround it’s still a workaround and not something you would understand yourself from just looking at the software.
maxB: I also have the exact same problem. My attempted workaround was to save the model by a new name for each design option, open a separate instance of the layout document for each option and relink the layout document to each separate model, and then copy all content of these layout documents into one big document. Still clumsy and would create a lot of extra work if a new scene would be needed for all design options.
I suggest having all the viewing information (camera position, visible layers, style, shadows, fog, active section cuts etc) controlled from Layout. Then there could be an eyedropper-like tool that lets you pick up these settings from one viewport and apply to another. The tool would have checkboxes somewhere to let the user control exactly what to copy, e.g. camera and active section but not layers.
By storing this information completely in the viewport and not referencing it from SketchUp I don’t see how any collisions would occur in a team environment.
Edit: Essentially the workflow would be the same as with a purpose model except the purpose model is instead integrated in LO in a way that makes sense to users.