feat(030): Dataset Lifecycle Workspace — Stats Bar, split-view, inline-edit, bulk actions

Backend:
- Add MetricItem, StatsCounts, ColumnDescriptionUpdate, MetricDescriptionUpdate DTOs
- Add metric_count to DatasetItem and DatasetDetailResponse
- Add server-side filtering via ?filter=unmapped|mapped|linked|all
- Add PUT endpoints for column/metric description inline-edit
- Add HTML stripping validation (max 2000 chars, plain text)
- Add WebSocket dataset.updated event on task completion
- RBAC: plugin:migration:WRITE for mutations, READ for reads
- 14 pytest tests (stats, filter, metrics, inline-edit, validation, 502/503)

Frontend:
- Rewrite +page.svelte as split-view orchestrator (387 lines)
- StatsBar.svelte — 4 metric tiles with aria-pressed, server-side filter dispatch
- DatasetList.svelte — card grid with progress bars, checkboxes, search
- DatasetPreview.svelte — presentational detail panel (container fetches)
- ColumnsTable.svelte — inline-edit with type chips, bold/italic fallback
- MetricsTable.svelte — inline-edit with expression hints
- 52 vitest tests across 7 test files
- i18n: 11 EN + 11 RU keys (with_mapping/с_маппингом)

Spec:
- 15 issues resolved (semantic label, filtering, endpoints, conflict, metric_count, RBAC, ID types, ports, i18n, validation, a11y, propagation, resize, perf, ownership)

Status: 14/14 backend tests pass, 52/52 frontend tests pass, build succeeds
Risk: Low — T051 browser validation pending (non-blocking)

backend/src/core/superset_client/_datasets.py

@@ -86,6 +86,22 @@ class SupersetDatasetsMixin:
                         "description": col.get("description", ""),
                     }
                 )
+            metrics = dataset.get("metrics", [])
+            metric_info = []
+            for metric in metrics:
+                metric_id = metric.get("id")
+                if metric_id is None:
+                    continue
+                metric_info.append(
+                    {
+                        "id": int(metric_id),
+                        "metric_name": metric.get("metric_name"),
+                        "expression": metric.get("expression"),
+                        "verbose_name": metric.get("verbose_name"),
+                        "description": metric.get("description", ""),
+                        "metric_type": metric.get("metric_type"),
+                    }
+                )
             linked_dashboards = []
             try:
                 related_objects = self.network.request(
@@ -139,6 +155,8 @@ class SupersetDatasetsMixin:
                 "description": dataset.get("description", ""),
                 "columns": column_info,
                 "column_count": len(column_info),
+                "metrics": metric_info,
+                "metric_count": len(metric_info),
                 "sql": sql,
                 "linked_dashboards": linked_dashboards,
                 "linked_dashboard_count": len(linked_dashboards),
@@ -147,7 +165,7 @@ class SupersetDatasetsMixin:
                 "changed_on": dataset.get("changed_on"),
             }
             app_logger.reflect(
-                f"Got dataset {dataset_id} with {len(column_info)} columns and {len(linked_dashboards)} linked dashboards",
+                f"Got dataset {dataset_id} with {len(column_info)} columns, {len(metric_info)} metrics, and {len(linked_dashboards)} linked dashboards",
                 extra={"src": "get_dataset_detail"},
             )
             return result
@@ -169,12 +187,13 @@ class SupersetDatasetsMixin:
     # @PURPOSE: Обновляет данные датасета по его ID.
     # @SIDE_EFFECT: Modifies resource in upstream Superset environment.
     # @RELATION: CALLS -> [APIClient]
-    def update_dataset(self, dataset_id: int, data: dict) -> dict:
+    def update_dataset(self, dataset_id: int, data: dict, override_columns: bool = False) -> dict:
         with belief_scope("SupersetClient.update_dataset", f"id={dataset_id}"):
             app_logger.info("[update_dataset][Enter] Updating dataset %s.", dataset_id)
+            endpoint = f"/dataset/{dataset_id}?override_columns={'true' if override_columns else 'false'}"
             response = self.network.request(
                 method="PUT",
-                endpoint=f"/dataset/{dataset_id}",
+                endpoint=endpoint,
                 data=json.dumps(data),
                 headers={"Content-Type": "application/json"},
             )

backend/src/core/task_manager/lifecycle.py

@@ -71,6 +71,7 @@ class JobLifecycle:
         self.persistence_service = persistence_service
         self.executor = executor
         self.loop = loop
+        self._dataset_subscribers: dict[str, list[asyncio.Queue]] = {}
     # #endregion __init__
 
     # #region create_task [C:4] [TYPE Function] [SEMANTICS task,create,persist,schedule]
@@ -190,6 +191,12 @@ class JobLifecycle:
                     "Task lifecycle reached persisted terminal state",
                     extra={"task_id": task_id, "status": str(task.status)},
                 )
+                # Broadcast dataset.updated for mapping and documentation tasks (FR-024, R4)
+                if task.plugin_id in ("dataset-mapper", "llm_documentation"):
+                    dataset_ids = task.params.get("dataset_ids") or [task.params.get("dataset_id")]
+                    env_id = task.params.get("env") or task.params.get("environment_id", "")
+                    if dataset_ids and env_id:
+                        self._broadcast_dataset_updated(env_id, dataset_ids)
     # #endregion _run_task
 
     # #region resolve_task [C:3] [TYPE Function] [SEMANTICS task,resolve,mapping,resume]
@@ -309,5 +316,37 @@ class JobLifecycle:
                 )
             self.graph.resolve_future(task_id, True)
     # #endregion resume_task_with_password
+
+    # #region subscribe_dataset_events [C:2] [TYPE Function] [SEMANTICS dataset,subscribe,events,websocket]
+    # @BRIEF Subscribe to dataset.updated events for a specific environment.
+    async def subscribe_dataset_events(self, env_id: str) -> asyncio.Queue:
+        queue = asyncio.Queue()
+        if env_id not in self._dataset_subscribers:
+            self._dataset_subscribers[env_id] = []
+        self._dataset_subscribers[env_id].append(queue)
+        return queue
+    # #endregion subscribe_dataset_events
+
+    # #region unsubscribe_dataset_events [C:2] [TYPE Function] [SEMANTICS dataset,unsubscribe,events]
+    # @BRIEF Unsubscribe from dataset.updated events for a specific environment.
+    def unsubscribe_dataset_events(self, env_id: str, queue: asyncio.Queue):
+        if env_id in self._dataset_subscribers:
+            if queue in self._dataset_subscribers[env_id]:
+                self._dataset_subscribers[env_id].remove(queue)
+            if not self._dataset_subscribers[env_id]:
+                del self._dataset_subscribers[env_id]
+    # #endregion unsubscribe_dataset_events
+
+    # #region _broadcast_dataset_updated [C:3] [TYPE Function] [SEMANTICS dataset,broadcast,event,websocket]
+    # @BRIEF Broadcast dataset.updated event to all subscribers for a given environment.
+    def _broadcast_dataset_updated(self, env_id: str, dataset_ids: list[int]):
+        event = {"type": "dataset.updated", "payload": {"env_id": env_id, "dataset_ids": dataset_ids}}
+        if env_id in self._dataset_subscribers:
+            for queue in self._dataset_subscribers[env_id]:
+                try:
+                    self.loop.call_soon_threadsafe(queue.put_nowait, event)
+                except Exception:
+                    pass
+    # #endregion _broadcast_dataset_updated
 # #endregion JobLifecycle
 # #endregion JobLifecycleModule

backend/src/core/task_manager/manager.py

@@ -318,5 +318,19 @@ class TaskManager:
             add_log_callback=self._add_log,
         )
     # #endregion resume_task_with_password
+
+    # ── Dataset event delegates to JobLifecycle ──
+
+    # #region subscribe_dataset_events [TYPE Function] [C:2]
+    # @BRIEF Subscribe to dataset.updated events for an environment.
+    async def subscribe_dataset_events(self, env_id: str) -> asyncio.Queue:
+        return await self.lifecycle.subscribe_dataset_events(env_id)
+    # #endregion subscribe_dataset_events
+
+    # #region unsubscribe_dataset_events [TYPE Function] [C:2]
+    # @BRIEF Unsubscribe from dataset.updated events.
+    def unsubscribe_dataset_events(self, env_id: str, queue: asyncio.Queue):
+        self.lifecycle.unsubscribe_dataset_events(env_id, queue)
+    # #endregion unsubscribe_dataset_events
 # #endregion TaskManager
 # #endregion TaskManagerModule