Make language tags consistent

[cmungall]

Half the OBI definitions have lang=en, the others are plain literals/strings

➜  ontology git:(master) ✗ grep IAO_0000115 obi-edit.owl | grep -c lang=
656
➜  ontology git:(master) ✗ grep IAO_0000115 obi-edit.owl | grep -vc lang=
702

I don't think there is OBO guidance but I think intra-ontology consistency is always good, it seems your intent is to be explicit, I think it would be clearly safe to auto-convert all definitions to lang=en

[turbomam]

Background from @jamesaoverton

• there used to be a plain RDF literal but now there are only language-tagged literals and datatype-tagged literals. A literal that has no explicit language or datatype tag is implicitly an xsd:string typed datatype literal
• adding a tag in Protege takes extra clicks so a lot of people don't do it
• Some ontologies may appear to have duplicate textual assertions like definitions or labels, but that can because the object appears in multiple language or datatype tagged format. Sometimes this can be attributed to the object having different tags in different imports. OWL considers "shark"@en and "shark"^^xsd:string different value.

[turbomam]

@cmungall is explicitly calling out definitions, but this is equally applicable to the objects of other triples (like rdfs:label). There's some acknowledgement of that in the PR discussion.

This work will need to be applied to inputs other than obi-edit.owl, like ROBOT templates. The bare use of AL column headers (without specifying the language code like @en) isn't a complete solution.

[turbomam]

Attendees in today's meeting agree with the content of this PR and appreciate the clever implementation. The way Dragon AI modified the tab characters make break the expectations of the ROBOT template format.

I and or @sebastianduesing will build upon it.

[turbomam]

we'll finish this PR on definitions and start another one for labels.