Question Dynamicaly culling content for multiple cameras (URP) - practical?

I'm in the middle of writing an occlusion culling system for my procedurally-generated dungeon, and it occurred to me that I might at some point want to support couch co-op, which would mean culling the scene twice from two viewpoints.

I know that there are opportunities to intervene in the rendering process prior to cameras rendering, but it would be useful to know if there are performance gotchas in terms of enabling/disabling meshrenderers, changing layers, or activating/deactivating game objects (eg triggering multiple regathers or recachings of the scene).

What would be the preferred method of achieving this?