Run child tables cascade from run.id

Migration ccb21eacf5d5 deletes rows whose run is gone, then adds foreign keys
with ON DELETE CASCADE from run_node, run_metric and run_artifact to run.id.
Deleting runs is now the one statement. A node or metric written after its
run was deleted is dropped with an info line. Migrations run with foreign keys
off, since a batch rebuild of run would otherwise cascade into every child row.

backend/fluksio/api/routes/runs.py

@@ -851,10 +851,10 @@ async def retry_run(run_id: str, request: Request, user: CurrentUser) -> Any:
 def _drop_runs(session: Session, chosen: Any) -> None:
     """A selection of runs and everything hanging off them.
 
-    The run tables carry a plain string ``run_id`` and no foreign key, so
-    nothing cascades on its own and these four statements are the whole of it.
-    ``flow_run``, ``metric_minute`` and ``engine_event`` stay — they are the
-    observability record and are pruned on their own window.
+    The run tables' foreign keys cascade, so deleting the runs takes their
+    nodes, metrics and artifact rows with them. ``flow_run``, ``metric_minute``
+    and ``engine_event`` stay — they are the observability record and are
+    pruned on their own window.
 
     Two things this costs, both deliberate. ``RunNode.outputs`` *is* the stage
     cache, so a later run loses hits these would have served. And a node
@@ -864,9 +864,6 @@ def _drop_runs(session: Session, chosen: Any) -> None:
     keeps whatever a ``run_artifact`` row or a live message still names, so
     dropping the rows is enough and the hourly sweep reclaims the blobs.
     """
-    session.execute(delete(RunNode).where(col(RunNode.run_id).in_(chosen)))
-    session.execute(delete(RunMetric).where(col(RunMetric.run_id).in_(chosen)))
-    session.execute(delete(RunArtifact).where(col(RunArtifact.run_id).in_(chosen)))
     session.execute(delete(Run).where(col(Run.id).in_(chosen)))