Address normalization issues

[VladimirAlexiev]

(Follow-up from #43)

Address has some fields that suffer from normalization issues

• cityName, countrySubDivisionName, countryName are perhaps redundant given the respective Id fields in Address
• The implicit relation between these Ids should be made explicit in a gazetteer, not repeated in every address.
  Otherwise it's possible for two addresses to hold inconsistent info.

Counter-arguments:

• UNCEFACT perhaps cannot assume the existence and use of a global gazetteer of all cities (although there are several: Geonames, Wikidata, OSM).
• If such assumption/choice cannot be made, then cityId has no global meaning, so two different UNCEFACT databases/messages can use them to refer to different datasets
• Eg cityName can be filled, but cityId can be missing: then cityName should stay in Address. Same for the other 2 levels

[VladimirAlexiev]

For the record: as of today, no changes are made to the ontology. The redundancy / potential inconsistency issue remains.
UNCEFACT has a geographic gazetteer: UNLOCODE (although it doesn't have all possible cities / inhabited places).

I understand the business issues behind the redundancy, so I won't plead reopening this issue.