Index/search decoded DocRef attachment data

[vmttn]

Fixes

Fixes arkhn/fhir-river#422, arkhn/jpaltime#20

Description

• Creates a dedicated index for documentReference attachment data at the initialization of hapi.
• Overrides controllers to index the decoded attachment data when manipulating documentReference resources.
• Implements searching on this index.

Technical details

• No hibernate since we simply cannot extend the entity without forking hapi and without affect on other resources.
• Uses the elastic java api.

Tests

1. docker-compose up -d potsgres elasticsearch kibana
2. add use_narrowing_interceptor: false in application.yaml:hapi.fhir
3. mvn jetty:run
4. Go to http://localhost:8080/resource?serverId=home&pretty=false&_summary=&resource=DocumentReference
5. Create resource ref with {"resourceType": "DocumentReference", "content": {"attachment": {"data": "Zm9vCg=="}}} ("Zm9vCg==" == base64("foo"))
6. Go to http://localhost:5601/app/discover and validate that the data has been correctly indexed
7. Search with : http://localhost:8080/fhir/DocumentReference?content=foo