refactor: SnarkableTree

[distractedm1nd]

/// Represents a tree that can be used to verifiably store and retrieve [`Account`]s.
/// The methods of this trait are NOT run in circuit: they are used to create verifiable inputs for the circuit.
/// This distinction is critical because the returned proofs must contain all information necessary to verify the operations.
pub trait SnarkableTree: Send + Sync {
    fn process_transaction(&mut self, transaction: Transaction) -> Result<Proof>;
    fn insert(&mut self, key: KeyHash, tx: Transaction) -> Result<InsertProof>;
    fn update(&mut self, key: KeyHash, tx: Transaction) -> Result<UpdateProof>;
    fn get(&self, key: KeyHash) -> Result<AccountResponse>;
}

This trait has changed a bit historically and is now big stupid. We should find a better way to model this. Specifically, it doesn't need to be a trait and KDT can implement the methods directly - with process_transaction and get being the only public methods