Updates permitted datatypes of sh:message and sh:description, and other properties previously permitting rdf:langString (Issues 723 & 724)

[HolgerKnublauch]

Also allowing rdf:dirLangString where rdf:langString was allowed.

Also allowing rdf:HTML at sh:message and sh:description.

Closes #723
Closes #724

• See this document rendered online here

[HolgerKnublauch]

The PR also includes a minor clean up to the SPARQL abstract, reflecting our recent generalization of the scope of SHACL beyond just validation.

Note that in the diagram I switched to "text" instead of the (too long) xsd:string or rdf:langString or rdf:dirLangString or rdf:HTML.

[TallTed]

Note that in the diagram I switched to "text" instead of the (too long) xsd:string or rdf:langString or rdf:dirLangString or rdf:HTML.

Perhaps we should define rdf:text as a supertype, having subtypes of xsd:string, rdf:langString, rdf:dirLangString, & rdf:HTML (and others?)...?

[HolgerKnublauch]

Introducing a supertype would require a new type of inference, and is certainly out of scope for the SHACL WG, esp if it would go into the rdf: namespace.

[TallTed]

I should have suggested sh:text, but I see your point. Still, that suggests that the "too long" string of optional type values should still be used, instead of "text".

[HolgerKnublauch]

@TallTed I have opened a follow-up PR - please approve if that addresses the remaining issues #756