Update to go-did v0.11

[reinkrul]

Changes:

• VerifiableCredential.IssuanceDate is now a required *time.Time. I had to add them where it wasn't present, and change it to a pointer where it was present.

Earlier, it included these changes, which we decided to revert since it leads to too many changes in the Nuts node:

• The REST API and go-leia credential store assume JSON-LD arrays are compacted. Changing this, although still valid JSON-LD and semantically the same credential, would break expected behavior of API users. So I introduced a CompactingVerifiableCredential and CompactingVerifiablePresentation which essentially do the same as go-did earlier did for JSON-LD types; compacting arrays on JSON marshalling. This is the majority of the change. This can be removed when the VCR v2 API and go-leia credential store is replaced with something else.
• SignedDocument.UnmarshalProofValue assumed there was only 1 proof which is never the case any more, since it's now always marshalled into an array (even if it has just one value). I extended the function to not burden the caller with checking the number of proofs, since I don't expect there multiple proofs at any time (although still supported).
• PEX paths now always index array properties (mostly verifiableCredential[0]).

[woutslakhorst]

The REST API and go-leia credential store assume JSON-LD arrays are compacted. Changing this, although still valid JSON-LD and semantically the same credential, would break expected behavior of API users. So I introduced a CompactingVerifiableCredential and CompactingVerifiablePresentation which essentially do the same as go-did earlier did for JSON-LD types; compacting arrays on JSON marshalling. This is the majority of the change. This can be removed when the VCR v2 API and go-leia credential store is replaced with something else.

Could you elaborate on this? In go-leia in examples/jsonld I can change the template (remove the array from children) and it will work exactly the same. Is it only an API problem then?

[gerardsn]

I've added a commit to fix a VC validator issue

[woutslakhorst]

It seems this only happens for the JSON stores, we can just change them to JSONLD stores and the problem is solved.

• change all JSON collection types to JSONLD collection types
• change to the correct index and search queries
• rename the index, then the index will be rebuild from the kv kackup

Lots of tests have to be updated though.

[woutslakhorst]

The need for this change started with simplifying the PEX submission validation. In order to do it without a lot of code, we could change the go-did library so it always marshals to arrays instead of objects when n=1. But this had some more consequences which are now being fixed by this PR. So in short: simplification lead to lib change lead to monkey patch...

Maybe it's better to implement the PEX stuff using the complex approach?

[reinkrul]

The need for this change started with simplifying the PEX submission validation. In order to do it without a lot of code, we could change the go-did library so it always marshals to arrays instead of objects when n=1. But this had some more consequences which are now being fixed by this PR. So in short: simplification lead to lib change lead to monkey patch...

Maybe it's better to implement the PEX stuff using the complex approach?

On the other hand, if upcoming VC versions don't compact arrays in JSON-LD, we need it sooner or later?

[gerardsn]

over complete review, but requesting changes mainly on last comment