Add shared nginx-proxy role; deploy Hermes in Docker; rename Paperclip

nginx-proxy (new role):
- Shared nginx + certbot under /opt/nginx, replacing the copies that
  lived inside the Paperclip compose project
- Apps publish via tasks_from: add_vhost (cert marker -> issue cert ->
  write vhost -> nginx -t -> reload); a vhost that fails nginx -t is
  rolled back
- Certs issued one per domain via Cloudflare DNS challenge; domains
  registered as marker files in cert-domains/; optional LE staging
- Stock nginx.conf plus conf.d (directory mounts, not single-file
  mounts); catch-all server drops unknown Host/SNI
- Containers renamed to nginx and certbot

hermes_docker (new role):
- Official Hermes image, host networking, gateway and dashboard bound
  to loopback and published through nginx-proxy
- Dashboard login via basic-auth provider, secrets from the vault
- Main model set from hermes_model (fantasmic endpoint, key from vault);
  config.yaml overrides are merged from a 0600 staged file that is
  removed afterwards, so the key never appears in a command line
- Data dir owner-only (HERMES_HOME_MODE=0700)

paperclip-docker:
- Remove nginx, certbot and docker-compose download tasks; deploy with
  community.docker.docker_compose_v2 (idempotent, remove_orphans)
- Stop passing the Cloudflare token and certbot email to the Paperclip
  container via env_file
- Bind Postgres to 127.0.0.1 (host networking exposed it on all
  interfaces); stop chowning pgdata, which the postgres image owns
- Handler recreates the paperclip service so .env changes apply
- Domain m-v.wdw-labs.cloud -> cranium-command.wdw-labs.cloud

Playbooks:
- Add day1_deploy_nginx_proxy.yml and day1_deploy_hermes_docker.yml
- deploy_paperclip_stack.yml: run nginx-proxy before paperclip-docker,
  CLI-based container checks (no Docker SDK dependency)
- Remove deploy-hermes.yml

Host vars: add nginx_proxy_certbot_email and nginx_proxy_certbot_staging
(staging stays on until the switch to real certs).

ansible/roles/hermes_docker/defaults/main.yml

@@ -0,0 +1,91 @@
+---
+hermes_image: nousresearch/hermes-agent:latest
+# missing = only pull if absent (no surprise upgrades on redeploy).
+# Set to "always" when you actually want to upgrade the image.
+hermes_pull_policy: missing
+
+hermes_container_name: hermes
+hermes_project_dir: /opt/hermes
+hermes_data_dir: "{{ hermes_project_dir }}/data"
+# The container's hermes user. The image's start-up hook chowns the data
+# volume to it; the role only needs the UID to stage files it can read.
+hermes_uid: 10000
+# config.yaml, .env and sessions live here and can hold API keys, so keep it
+# owner-only. Passed to the container as HERMES_HOME_MODE (see compose).
+hermes_data_dir_mode: "0700"
+hermes_secrets_env_file: "{{ hermes_project_dir }}/hermes-secrets.env"
+
+# Same network namespace as Paperclip and nginx
+hermes_network_mode: host
+hermes_memory_limit: 8G
+hermes_cpu_limit: "4.0"
+# Playwright/Chromium needs shared memory (upstream troubleshooting note)
+hermes_shm_size: 1g
+
+# --- Gateway / API server (loopback only; Paperclip reaches it locally) ---
+hermes_gateway_port: 8642
+hermes_api_server_enabled: true
+hermes_api_server_host: 127.0.0.1
+# Leave empty to run keyless on loopback. If you set one, Paperclip (or any
+# other client) must send it as a bearer token.
+# e.g. hermes_api_server_key: "{{ vault_hermes_api_server_key }}"
+hermes_api_server_key: ""
+
+# --- Dashboard (bound to loopback, published through nginx) ---
+hermes_dashboard_enabled: true
+hermes_dashboard_host: 127.0.0.1
+hermes_dashboard_port: 9119
+hermes_domain: haunted-mansion.wdw-labs.cloud
+# Publish the dashboard through the nginx-proxy role (cert + vhost + reload).
+# Requires day1_deploy_nginx_proxy.yml to have run first.
+hermes_publish_vhost: true
+hermes_public_url: "https://{{ hermes_domain }}"
+hermes_dashboard_username: admin
+
+# Vault variables this role requires (see README block at bottom of file)
+hermes_required_vault_vars:
+  - vault_hermes_dashboard_password_hash
+  - vault_hermes_dashboard_secret
+
+# --- Preflight ---
+hermes_preflight_ports:
+  - "{{ hermes_gateway_port }}"
+  - "{{ hermes_dashboard_port }}"
+hermes_wait_timeout: 120
+
+# --- config.yaml overrides (deep-merged, only the keys listed here) ---
+# nginx runs with host networking, so it reaches the dashboard from loopback.
+# Model settings are NOT managed by default (your wizard-written config stays
+# untouched). To manage them, add a model block, for example:
+#   model:
+#     provider: custom
+#     model: <served-model-name>
+#     base_url: http://10.1.71.130:8000/v1
+#     api_key: "{{ vault_vllm_api_key }}"
+hermes_config_overrides:
+  dashboard:
+    trusted_proxies:
+      - "127.0.0.1"
+
+# Main model (written to config.yaml under model:). Leave empty to manage it
+# by hand with `hermes model` or the dashboard. Example:
+#   hermes_model:
+#     provider: custom
+#     default: <served model name from the endpoint's /v1/models>
+#     base_url: http://fantasmic.wdw-labs.cloud:8000/v1
+#     api_key: "{{ vault_fantasmic_api_key }}"
+#     # context_length: 65536   # only if Hermes complains about the context window
+# The api_key lands in config.yaml in plain text (owner-only, see hermes_data_dir_mode).
+hermes_model: {}
+
+# What the merge task actually applies: base overrides + the model block.
+hermes_effective_overrides: >-
+  {{ hermes_config_overrides
+     | combine({'model': hermes_model} if (hermes_model | length > 0) else {}, recursive=True) }}
+
+# Required vault vars:
+#   vault_hermes_dashboard_password_hash  - output of hash_password() (see below)
+#   vault_hermes_dashboard_secret         - openssl rand -base64 32
+# Generate the hash with:
+#   docker exec -it -w /opt/hermes hermes /opt/hermes/.venv/bin/python -c \
+#     "from plugins.dashboard_auth.basic import hash_password; print(hash_password('PW'))"