Introduce dedicated types for DID and handle Lexicon string formats

[str4d]

No description provided.

[str4d]

I've opened this PR to get review on the general approach, as well as the types that were simpler to implement. I am also working locally on:

• A Datetime type that wraps chrono::DateTime.
  • This one is annoying because Lexicon uses the intersection of RFC 3339 and ISO 8601, and chrono can't parse the latter).
• An Nsid type that wraps a String.
• An AtUri type that contains an AtIdentifier, an Nsid, and a RecordKey.
• A Uri type that might wrap url::Url.
  • I'm not yet sure if this will be valid. The at-uri format explicitly says it is not compatible with the WHAT-WG URL specification, which is what the url crate implements. The uri format seems like it is intended to be more general, but it does say it can include AT URIs, so that probably means this won't work.

[sugyan]

I've read the documentation.
Lexicon requires the intersection of RFC 3339 and ISO 8601, and chrono supports RFC 3339 with parse_from_rfc3339, but just because it can parse it does not mean it conforms to ISO 8601. So, if we want to implement it correctly, we need to exclude the ones that are not correct as ISO 8601. Am I correct in my understanding that this is the case?

If this library is more permissive than the spec (though it would need documentation), then all input for RFC 3339 format should be fine if parse succeeds, and the output in to_rfc3339 should meet Lexicon's requirements. I suspect that would work well enough, but what do you think?

[str4d]

I've read the documentation. Lexicon requires the intersection of RFC 3339 and ISO 8601, and chrono supports RFC 3339 with parse_from_rfc3339, but just because it can parse it does not mean it conforms to ISO 8601. So, if we want to implement it correctly, we need to exclude the ones that are not correct as ISO 8601. Am I correct in my understanding that this is the case?

Correct, that is indeed the case for parsing. I think this is primarily things like not parsing the T and Z as case-insensitive, but I'm trying to figure out a nice way to handle them that isn't just double-parsing.

If this library is more permissive than the spec (though it would need documentation), then all input for RFC 3339 format should be fine if parse succeeds, and the output in to_rfc3339 should meet Lexicon's requirements. I suspect that would work well enough, but what do you think?

For serialization, I'll be using DateTime::.to_rfc3339_opts(SecondsFormat::Micros, true), which should generate a timestamp string that is ISO 8601 compliant.

[str4d]

Rebased on main to fix merge conflict with #102 and address review comment.

[sugyan]

Thank you very much.
I will merge this once, can you please proceed with another pull-request for DateTime, Uri, etc.?