Fix stale sftp IP, ansible.cfg keys, drop dead shutdown playbook

- sftp IP 192.168.99.68 in CLAUDE.md and server README (DMZ move)
- ansible.cfg: ansible_python_interpreter is a host var, moved to
  group_vars/all.yml; [ssh_connections] -> [ssh_connection] so
  pipelining actually applies
- remove playbooks/shutdown.yml (targeted nonexistent k3s_cluster group)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

ansible/ansible.cfg

@@ -1,9 +1,8 @@
 [defaults]
 nocows = 1
 host_key_checking = false
-inventory = ./hosts.ini 
-ansible_python_interpreter  = /usr/bin/python3
+inventory = ./hosts.ini
 vault_password_file = ~/.ansible/vault_pass
 
-[ssh_connections]
+[ssh_connection]
 pipelining = true

ansible/group_vars/all.yml

@@ -2,6 +2,7 @@
 # connection settings (identical for all hosts)
 ansible_user: sbxadmin
 ansible_port: 22
+ansible_python_interpreter: /usr/bin/python3
 
 # generic settings
 main_username: sbxadmin

ansible/playbooks/shutdown.yml

@@ -1,9 +0,0 @@
----
-- name: Shutdown k3s_cluster
-  hosts: k3s_cluster
-  gather_facts: true
-  tasks:
-    - name: Shutdown the nodes (and wait one 1 min)
-      become: true
-      community.general.shutdown:
-        delay: 60