Refactoring of the code maintaining ClusterNodes (#6466)

* Using Arc<str> for node ids.

* ClusterMember as a member in ClusterNode

* Marked ready_members_watcher as for test only

* Moved methods to deprecate in their own block

* Switched to chitchat 0.11

* Remove delegating methods from ClusterNode that duplicate ClusterMember fields via Deref

Methods removed: is_ready, grpc_advertise_addr, ingester_status, indexing_tasks, indexing_capacity.
All call sites updated to access fields directly through the existing Deref<Target = ClusterMember>.
Note: indexing_capacity() was an alias for the indexing_cpu_capacity field.

* Remove ready_members_rx watch channel and spawn_ready_members_task

The deprecated watch channel and background task maintained a redundant copy
of the ready member list. The test-only helpers (ready_members, wait_for_ready_members)
are reimplemented on top of the existing change_stream(), which already provides
consistent snapshots plus incremental updates under the write lock.
The test_single_node_cluster_readiness test is updated accordingly.

* Moving util methods from Node to Member

* Removing methods to build convert NodeId

* Cargo fmt + CR comment

---------

Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>