Optimize node processing

[MondayHopscotch]

Currently, the tree starts from the root and works its way down the nodes every frame. This potentially causes a lot of overhead, especially in larger trees.

Current planned approach:

• As process() is called on nodes, they can add themselves to the Executor's activeNodes list.
• If there are any activeNodes when the Executor's process() is called, it calls process() on only those nodes in depth order
• If any node reports a change (We'll have to figure out what this means), we clear out any activeNodes with depth > the depth of the changed node
• If for any reason there are no activeNodes, then the Executor starts from the root