Fix ES server status detection by the status servlet

Steps to reproduce:

• on 9.10 (or 10.1-SNAPSHOT), configure a Nuxeo to hit a ES server by adding to the nuxeo.conf, for instance:

elasticsearch.enabled=true
elasticsearch.client=RestClient
elasticsearch.addressList=localhost:9200
elasticsearch.clusterName=mycluster
elasticsearch.indexName=myindex
elasticsearch.indexNumberOfReplicas=0

• start Nuxeo
• check status servlet response at http://localhost:8080/nuxeo/runningstatus, it should return:

{"runtimeStatus":"ok","elasticSearchStatus":"ok","ldapDirectories":"ok","repositoryStatus":"ok"}

• check that search works alright in UI
• stop the ES server
• check that the search does not work ok anymore in the UI
• check status servlet response: it has not changed even if it should now state (eventually after the 20s - default cache value on probes)

{"runtimeStatus":"ok","elasticSearchStatus":"failed","ldapDirectories":"ok","repositoryStatus":"ok"}

When restarting the ES server, the search is working alright again (ES service reconnection works ok).