fix(node decomission): stop a decommissioned node by default

[yarongilor]

A decommissioned node currently left is_running() after its
status changed to "decommissioned". thus it is unexpectedly left
as part of cluster nodes, like in cluster.nodelist().
This should not be the default, but rather to stop() the node
after decommission successfully completed (as it is in SCT).
This fix requires a corresponding dtest PR to adjust all calls of
node.decommission() in order to remove 14 un-needed calls to node.stop()
or to set the new flag of stop_node=False needed for raft testing.
This fix is a followup of:
https://github.com/scylladb/scylla-dtest/pull/4767#discussion_r1731340260

refs: https://github.com/scylladb/scylla-dtest/pull/4767#discussion_r1731340260

There are also 11 occurrences of nodetool("decommission") to be adjusted/optimized:

$ grep -Eri '\.nodetool\(.*decommission' --include \*.py .
./manager_backup_tests.py:        node3.nodetool("decommission")
./compaction_additional_test.py:                node2.nodetool("decommission")
./manager_restore_tests.py:        node3.nodetool("decommission")
./manager_restore_tests.py:        node3.nodetool("decommission")
./manager_restore_tests.py:            node.nodetool("decommission")
./update_cluster_layout_tests.py:        node1.nodetool("decommission")
./update_cluster_layout_tests.py:        node3.nodetool("decommission", capture_output=False, wait=False)
./update_cluster_layout_tests.py:        node3.nodetool("decommission", capture_output=False, wait=False)
./update_cluster_layout_tests.py:        node3.nodetool("decommission", capture_output=False, wait=False)
./nodetool_additional_test.py:        node2.nodetool("decommission")
./topology_test.py:            out, err = node.nodetool("decommission")