Entity name matching should ignore diacritics, accents and arabic hamza even without fulltext index

This can be achieved by introducing a new technical field that stores normalized version of the names (title + alternative names and transliterations if any).

Queries should happen on this field using the same normalization function.