Encoding to canonical format (via RFC 8785)

[jamesaimonetti]

https://www.rfc-editor.org/rfc/rfc8785#name-serialization-of-strings stipulates encoding of \u000F MUST be \u000f (more generally \uhhhh vs \uHHHH).

Any interest in an encoding option to "canonical" form? Helpful for HMACs and the like.

[davisp]

I'd accept a [canonical] option patch. The lower case hex stuff is fairly trivial, but the sorted UTF-16 encoded keys is a bit of a head scratcher. Perhaps do that in Erlang before the ErlJSON object is passed to the encoder?