OCI IPLD Transition

[10d9e]

Phase 2 of IPCR will be refactoring the existing IPFS/unixfs registry data model into an IPLD data model (OCI ADL, codec, etc). In general this transition should include the following:

• high level design artifact; markdown document that describes the new data model and what the transition from the current OCI model will look like. (see second half of Estuary/IPCR presentation: https://docs.google.com/presentation/d/1_Y3xmtEaVtCAmSk-ezEVW0CiRNv9xGClex3tRZ2kVzA/edit?usp=sharing)
• remove/address mapping between SHA256 and IPFS CIDS that currently takes place. Because the "hash sum" of an IPFS file is the root of an underlying merkle dag, it doesn't deterministically map to the traditional sha256 hash sum. (ie. shasum -a 256 <file>). Working this into the ADL/Codec would be the most ideal. The alternative would be to have replace the sha sums with the CID sums in the registry itself... not impossible, and perhaps more "IPLD-ish", but further reaching into the registry codebase.
• ideally has an interface to do IPLD querying/traversal by name/version, for example nginx:v1.2
• as with version 1, this version of the registry must work with the native Docker client - all of the integration should happen in the registry itself, allowing for a transparent experience to the end user

As this develops, it is also a good idea to document best practices in generating new hashlinked structures with IPLD for Estuary consumption. (ie. a github template project)