cephadm: warn when orch actions are deferred on unmanaged services

When a service is unmanaged, ceph orch apply / daemon restart|redeploy|reconfig still schedule work, but reconciliation no-ops until the service is managed again. Append a clear NOTE to those command responses, document the set-managed → wait → set-unmanaged workflow, and clarify the bare osd special case. Tests updated accordingly.

Fixes: https://tracker.ceph.com/issues/77593
Signed-off-by: Yael Azulay <yazulay@redhat.com>

Co-authored-by: Cursor <cursoragent@cursor.com>