manage elasticsearch datasets

[jum-s]

today, images of elasticsearch and entities-serch-engines are deployed from the docker compose file.
This decision has impact on the devlopement process : to ease devlopment, a container backup via volumes (more info) must to be created since all containers may be removed in a development environment (hence everytime de developer has to reimport dataset via dump_wikidata_subset script
Another solution from maxlath could be to make available remote eleasticsearch instance from the alt-prod server. Which would ease the docker setup of the devlopment environment, but might clash once we would like to deploy inventaire in production via docker.