refactor(objstore): extract MinIO into dedicated role 'web-app-minio' and adjust AI role

• Rename ports: web-app-ai_minio_* → web-app-minio_* in group_vars

• Remove MinIO from web-app-ai (service, volumes, ENV)

• Add new role web-app-minio (config, tasks, compose, env, vars) incl. front-proxy matrix

• AI role: front-proxy loop via matrix; unify domain/port vars (OPENWEBUI/Flowise *_PORT_PUBLIC/_PORT_INTERNAL, *_DOMAIN)

• Update compose templates accordingly

Ref: https://chatgpt.com/share/68d15cb8-cf18-800f-b853-78962f751f81

roles/web-app-ai/templates/env.j2

@@ -9,7 +9,7 @@ LITELLM_MASTER_KEY=dummy-key
 LITELLM_CONFIG=/etc/litellm/config.yaml
 
 # Flowise
-PORT={{ AI_FLOWISE_INTERNAL_PORT }}
+PORT={{ AI_FLOWISE_PORT_INTERNAL }}
 FLOWISE_USERNAME=admin
 FLOWISE_PASSWORD=admin
 DATABASE_PATH=/root/.flowise
@@ -19,7 +19,3 @@ FLOWISE_FILE_STORAGE_PATH=/root/.flowise/storage
 QDRANT_URL={{ AI_QDRANT_INTERNAL_URL }}
 OPENAI_API_BASE={{ AI_LITELLM_INTERNAL_URL }}/v1
 OPENAI_API_KEY=dummy-key
-
-# MINIO
-MINIO_ROOT_USER=admin
-MINIO_ROOT_PASSWORD=adminadmin