Fix JCS encoding

[clehner]

There is a bug in the serde_jcs crate in the encoding of property names for sorting: https://github.com/l1h3r/serde_jcs/issues/1

ssi uses serde_jcs:

1. in src/ldp.rs: JcsTezosSignature2021 (#237), for encoding linked data documents for signing.
2. in src/eip712.rs: in a test function, used to generate a test vector for Ethereum EIP712 Signature 2021
3. in src/jsonld.rs: for serializing strings and numbers

• 2 I do not think is affected by the bug, because the document uses only ASCII property names, which sort the same in UTF-8 as in UTF-16 (RFC 8785 (JCS) uses UTF-16 and notes this at the end of Section 3.2.3).
• 1 will be affected by this bug if documents use non-ASCII property names.
• 3 is not affected by the bug, since it is used for serializing only strings and numbers, not objects or arrays, so does not involve string comparisons that affect the output.