DAGJson support for Ipld

[austinabell]

Summary of changes
Changes introduced in this pull request:

• Useful going forward for selectors and testing, implementation is not efficient since the serialization and deserialization requires some theoretically unnecessary clones/copies, but it doesn't matter because this is just JSON and I don't want to fork serde_json or reimplement
  • Added json support for Cids with this (for obvious reasons) but did not for anything outside of IPLD because I don't think it'll be necessary yet
• Updates serialization tests to use Cid json and cleaned up a few other small things

Also added the ipld! macro for fun (makes things a lot easier)

Reference issue to close (if applicable)

Closes

Other information and links

[austinabell]

So pr is blocked by breaking changes in rust-fil-proofs being updated, but also could be some functional changes depending on what happens with ipld/specs#259

Isn't a big rush to get this in so can probably wait for that to resolve

[ec2]

Feels weird having a doc test for how to use an Enum... Just adds to test compilation and running without much value. Your documentation of the macro is pretty good already

[austinabell]

Made more sense to me to document how to construct each variant instead of doing all on the Ipld enum. These are pretty lightweight but I can add no_run to them or remove, just nice to have examples for each variant for practical use imo.

[ec2]

This is so janky its funny. But agree we dont need to fix this yet haha

[austinabell]

I mean, it's not THAT janky, not like it's a common concept to be able to go from a generalized type to a specific one without a buffer usually. This is correct, just not efficient

[ec2]

Definitely correct, not questioning. I just think its funny how we have to go to CBOR first then deserialize. Anyways, nothing needs to be fixed here.