Amino can be removed from the proofs. This will affect the IavlSpec in ICS-23 since the prefix would change. Think it's worth it to purge iAVL of amino though.

Think #276 should be merged after this PR, and IavlSpec needs to be changed in https://github.com/confio/ics23

Please confirm @ethanfrey

As to ics23 proofs, we allow a spec to length-prefix data many ways. Ideally the output of binary.Varint is compatible with one of the supported methods. If so, we just update the spec along with the new proofs. If this is different, please make a PR to add this to ics23 (we need it in proto and 3 languages) or use a supported format, or wait til I have bandwidth for this.

It looks like binary.Varint is protobuf compatible:

The varint functions encode and decode single integer values using a variable-length encoding; smaller values require fewer bytes. For a specification, see https://developers.google.com/protocol-buffers/docs/encoding. 

So this shouldn't even be an issue for IavlSpec (but please test)

I would also say that once #276 is merged and deeply integrated (in a separate PR?) to access the store rather than translating exisiting proofs, you may wish to remove the custom amino-style proof format. How are RangeProofs serialized? Maybe not in this module, but in the sdk they are only serialized using amino before hitting the wire on sdk queries.

Following discussions in a call last week with @AdityaSripal and @alexanderbez, we'll keep the current proofs along with the new proofs, but try to migrate them to Protobuf if we are able to keep byte-compatibility.

I'll do that in a separate PR, so this PR is good to review and merge.