[MMF-2168] Hazelcast adjustments and documentation for DCE Docker Images

- https://www.elastic.co/guide/en/elasticsearch/reference/7.x/breaking-changes-7.0.html#_discovery_configuration_is_required_in_production

- Orchestrator upgrade was required as by default it was setting these properties

- add should minimum match eq 1 to user index queries
ES 7.X changed behaviour in case filter query with bool it defaults to '0'
https://www.elastic.co/guide/en/elasticsearch/reference/7.x/breaking-changes-7.0.html#_the_filter_context_has_been_removed

- fix issue index routing param
ES 7.X helped discover this bug as new setting has been auto configured which is 'index.number_of_routing_shards'.
This has changed how documents are distributed across shards depending on how many shards the index has.

Without that change issues docs has been incorrectly routed to the same shard hash as projects and it worked no matter what routing key you used projectUuid or auth_projectUuid.

- update ngram and edge_ngram names to match with es 7.x
nGram and edgeNgram has been deprecated in favour of ngram and edge_ngram
https://www.elastic.co/guide/en/elasticsearch/reference/current/breaking-changes-7.0.html#deprecated-ngram-edgengram-token-filter-cannot-be-used

- remove `_all : enabled` usage from UT
This field was already deprecated in 6.X, now it has been removed.
https://www.elastic.co/guide/en/elasticsearch/reference/current/breaking-changes-7.0.html#all-meta-field-removed

- add Elasticsearch High Level REST client dependency

- use sonar.search.port for ES HTTP

- main process use ES Rest client to check ES status

- sonar.cluster.search.hosts has HTTP ports on APP nodes
also sonar.search.port and sonar.search.host MUST be configured on each Search node with the host and HTTP port of the current node

- use Elasticsearch high level rest client

- use in EsTester

- use as primary es client

- use indices api to get all indices name instead of cluster api

- use cluster health api to check cluster state

- support raw requests for 'nodes/_stats' and '_cluster/stats'

- support raw requests for 'indices/_stats'

- leave netty4plugin as testCompile dependency it is used in UTs

- all ES non-test calls go through EsClient class

- add rest client ES profiling

deprecated "discovery.zen.ping.unicast.hosts" has been removed, "discovery.seed_hosts" should be used instead
"discovery.zen.minimum_master_nodes" is ignored now: stop setting it and drop sonar property "sonar.search.minimumMasterNodes"

ES 7.5.1 refuses to start without this option

node.store.allow_mmapfs was deprecated and is now dropped in 7.5.1

Elastic introduced "system" JVM settings in 7.X:
 * "system" settings are set from Java code (class SystemJvmOptions)
 * which means these JVM settings are not defined in "jvm.options" file anymore

and consistently valid adresses

since this code is interrupting other threads, this ensures concurrent threads attempting to run this code will never interrupt each other
leading to unpredictable and unreproducible behavior

this will be helpful as sonar-main has several concurrency testing UTs

this avoid having multiple thread call the stope finalization concurrently and killing each other, which is unpredictable and very hard to reproduce behavior

and for the sake of clean logs, indicate CE is operational

this prevents concurrent threads to shutdown each other
this removes a test on NodeLifecycle.getState() which could be unthreadsafe and keep all lifeCycle logic in NodeLifecycle class

this allows to keep stop code behing lifecycle transition test
which prevents multiple stop sequence to be initiated in parallel from different thread
which may interrupt each other and create unstable behaviors (with cryptic non reproducible errors)

display Thread name more often
display call stack at trace level

otherwise, two hard stopping threads may interrupt each other and SQ could not stop at all

rethrowing is effectively useless as no called will catch the exception
in practice, it is caught by the default exception handle of technical thread involved in scheduler SQ start/stop and logged to default logger
this is just noise as the error has been logged by ManagedProcessHandler's start method anyway.

same applies to StopWatcher thread