Support JSON5 deserialization

[CompeyDev]

This PR implements JSON5 deserialization (and serialization too, for the sake of completeness) for the serde and net builtin libraries.

TODO:

• Implementation
  • @lune/serde
    • encode
    • decode
  • @lune/net
    • jsonDecode
• Tests & Documentation
  • Document new json5 union variant
  • Include tests

Closes #141.

[filiptibell]

It seems this PR conflates two different standards (json and json5) in a couple places. In previous discussion on Discord I also rejected the proposal to add json5, with the two main points being:

• The serde_json5 crate is for all intents and purposes unmaintained, and is a fork of an even older, even less maintained crate with many pending issues
• Any changes they have made/accepted since the initial fork release, seem to not have been published

A better option here would be "json with comments", using something like https://crates.io/crates/jsonc-parser. This is human-friendly JSON that doesn't try to include lots of extra ECMAScript syntax.

[CompeyDev]

It seems this PR conflates two different standards (json and json5) in a couple places.

Could you elaborate? JSON is technically valid JSON5, and so encoding is the only place I've used them interchangeably.

[CompeyDev]

Also, if you could reopen the PR, I guess I could replace the existing changes to make use of JSONC instead, considering the changes were seemingly trivial.

[filiptibell]

Could you elaborate? JSON is technically valid JSON5, and so encoding is the only place I've used them interchangeably.

For net.jsonDecode, using json5 brings a performance penalty, it's also not really used for sending over the network and I don't think we should encourage that. The json5 spec repo also says the same.
Serializing both json and json5 as plain json also breaks some expectations - keys should be bare strings whenever possible with json5 according to their spec tests, not quoted like json. I'm not sure if serde_json5 does this or not though but if it doesn't that's another minus point for it

Also, if you could reopen the PR, I guess I could replace the existing changes to make use of JSONC instead, considering the changes were seemingly trivial.

I would like to keep this one closed so that it is searchable/linkable in the future, but feel free to make another PR

[CompeyDev]

Thanks for the clarification, guess I'll create a new PR in the future once I've finished the ones I'm currently working on.