The term bibliographic number is easily confused with "inventory number" and any local or old internal record numbers that migrated MARC may have. 

I think the most accurate long form  representation is "Koha internal record number", as the biblionumber has nothing at all to do with MARC, and wouldn't necessarily be carried over to another system if the record was moved (not that anybody would *leave* Koha, amirite).

With that in mind, I don't see why we would need to drop biblionumber at all - it references exactly what it is, is specific to Koha, and can be understood and trained in that context, fully independent of MARC.