api: Make the api consistent

[evan-forbes]

When we push data to the tree, we are using prefixed data, but when we verify proofs, we are using non-prefixed data. Whatever the API ends up being, we should probably make it consistent.

This also applies to creating and verifying proofs. When creating proofs, we have two different functions, one for creating a ranged proof, and another for creating a proof to a single leaf. When verifying proofs, we are only using a single function to verify proofs of a single leaf or a range of leaves. Whatever we choose, we should probably make this consistent as well.

relevant comment #58 (comment)

related to #55

[staheri14]

When we push data to the tree, we are using prefixed data, but when we verify proofs, we are using non-prefixed data. Whatever the API ends up being, we should probably make it consistent.

A quick note on this point: in the current state of the nmt, the VerifyNamespace accepts namespaced leaves (which is consistent with the Push in terms of accepting namespaced data)