Manage a pool of clients [API-2055]

[zpqrtbnk]

Provide a way to manage a pool of clients, which would remain connected to the cluster, so instead of having to manage their own caching of a client, users can rely on this pool.

Question: why do this and not use one singleton client?

[ssahadevan]

Note from Xin: When we need to launch many threads/tasks to fetch key/value’s from hazelcast, we need to use multiple clients managed by a pool. Current API has a method getAllClients(), but it does not provide pool-like mechanisms such as takeOne() and returnToPool()

[github-actions[bot]]

Internal Jira issue: API-2055