Implement Batch Rendering by enabling users to 'flatten' a MeshNode tree

I would like to implement some form of batch rendering, where the vertex data for multiple meshes/models are stored contiguously in a single VAO, instead of a 1:1 Mesh:VAO relationship.

However, I do like the logical clarity of the MeshNode tree and the way resources currently work (in the upcoming Engine-Overhaul branch).

One solution could be to have a process where a whole MeshNode tree can be flattened into a single MeshNode with one mesh. We could then say that MeshNode now also points to its 'flat' representation as well as its children. When we attempt to draw a meshnode with a 'flat' representation, we would now just draw this flat version, instead of descending the tree and drawing all its children individually.

We would also likely want some mechanism that unloads children of a node that has been flattened (from the GPU), as these will never be drawn by OpenGL directly, and are now just logical representations?