I'm trying to have objects in a game be selectable only from a top down orthographic camera view. I think having physics colliders might be overkill for that - especially because this only needs to be in 2D and not 3D. Is there another method that might be better, such as something like a renderer "raycasts", but perhaps even more performant as it is just from the orthographic 2D top down view
If you build your "top down" view on the XY plane, you can use 2D colliders and
This is about as efficient as you will get. A 2D collider is using exactly the same math you'd need to write in your C# code to test for a hit anyway. And if you're using Kinematic bodies on a physics layer that doesn't interact with anything, they won't burn any cycles on unneeded collision detection and resolution.