fix: refresh app state on task completion for both success and failure

Previously, app details were only refreshed on successful task completion via UpdateAppDataHandler. When tasks failed (e.g., blueprint command errors), the state machine jumped directly to SetFailed, leaving stale container IDs in the app state.

Now TaskCompletionHandler always calls inspect_app() and update_app() before completing, ensuring app state reflects current Docker container information regardless of task outcome.

Closes #605

Author: stmh

scotty/src/docker/state_machine_handlers/task_completion_handler.rs

@@ -4,7 +4,7 @@ use scotty_core::{notification_types::Message, tasks::task_details::State};
 use tokio::sync::RwLock;
 use tracing::instrument;
 
-use crate::state_machine::StateHandler;
+use crate::{docker::find_apps::inspect_app, state_machine::StateHandler};
 
 use super::context::Context;
 
@@ -78,23 +78,29 @@ where
             CompletionType::Failure => (State::Failed, "Operation failed for", true),
         };
 
-        // Use the shared helper - single source of truth for task completion
-        let app_name = context.read().await.app_data.name.clone();
-        let status_msg = format!("{} operation for app '{}'", status_msg_prefix, app_name);
+        // Refresh app state to get current Docker container info
+        {
+            let ctx = context.read().await;
+            let docker_compose_path = std::path::PathBuf::from(&ctx.app_data.docker_compose_path);
 
-        context
-            .read()
-            .await
-            .complete_task(target_state, status_msg, use_error_status)
-            .await;
+            let app_data = inspect_app(&ctx.app_state, &docker_compose_path).await?;
+            ctx.app_state.apps.update_app(app_data).await?;
+
+            // Use the shared helper - single source of truth for task completion
+            let app_name = ctx.app_data.name.clone();
+            let status_msg = format!("{} operation for app '{}'", status_msg_prefix, app_name);
+
+            ctx.complete_task(target_state, status_msg, use_error_status)
+                .await;
+        } // Drop ctx read lock here
 
         // Send notifications in a dedicated thread (for both success and failure)
         if self.notification.is_some() {
             tokio::spawn({
                 let notification = self.notification.clone();
                 let completion_type = self.completion_type;
+                let context = context.clone();
                 async move {
-                    let context = context.clone();
                     let context = context.read().await;
 
                     if let (Some(app_settings), Some(notification)) =