Save enrichment data to persistent storage

Save both the raw data and formatted data to a persistent store
The data storage should respect the following usage requirements:

• Used for search of documents and aggregation
• Be up-to-date: if the field used to create it has changed, it should be re-calculated
  • in the case the annotation was human, we should not lose it. Consider a way to store the manual annotation and the field that was used to create it.
• Have normalised data to a specific kind, but store all data returned by the service.
• allow for re-normalisation:
  • if we decide that now we want to accept all tags that have 60% probability or more, and not only those above 80%, no need to recalculate all