fix(roles): detect new image on podman pull and restart services

Same bug as immich (556fae8): 'Writing manifest' goes to stderr, so
changed_when never matched and the restart handlers never fired after
image upgrades. Compare local image IDs before/after pull instead, and
notify the restart handler only on a genuinely new image.

Applies to fdroid, gitea, metabase, ntfy, unifi, uptime-kuma.

roles/ntfy/tasks/main.yml

@@ -37,13 +37,22 @@
     mode: "0644"
   notify: Restart ntfy
 
-- name: Pull ntfy container image
-  ansible.builtin.command: "podman pull {{ ntfy_image }}:{{ ntfy_version }}"
-  changed_when: pull_result.stdout is search('Writing manifest')
-  register: pull_result
+- name: Get current ntfy image ID
+  ansible.builtin.command: "podman image inspect --format {% raw %}{{.Id}}{% endraw %} {{ ntfy_image }}:{{ ntfy_version }}"
+  register: ntfy_image_id_before
+  failed_when: false # image may not exist yet on first deploy
+  changed_when: false
   become: false
   become_user: "{{ ansible_user }}"
 
+- name: Pull ntfy container image
+  ansible.builtin.command: "podman pull --quiet {{ ntfy_image }}:{{ ntfy_version }}"
+  register: pull_result
+  changed_when: pull_result.stdout != ntfy_image_id_before.stdout
+  become: false
+  become_user: "{{ ansible_user }}"
+  notify: Restart ntfy
+
 - name: Deploy Kubernetes YAML for ntfy
   ansible.builtin.template:
     src: ntfy.yaml.j2