Use a single memory block

[0undefined]

Make a single allocated memory block for the whole tree.

Pros:

• More performant memory layout in regards to memory prefetching
• This means (probably) faster linear iteration of the tree
• Fewer memory allocations on insertions
• Allows for the possibility of creating a tree from a known/static dataset

Cons:

• More difficult memory management
• Increases unused memory (can possibly be mitigated if we allow an argument for how many items are expected to be in the tree)

Strategies

1. Place everything in a blob pointing to a sequence of {node, elem [2*t]}
2. Have seperate arrays for nodes and elements respectively

1 seems bothersome but easy, however 2 might be easier to implement with the current state of things. Also 2 is probably nicer to work with.