fix(preprocess): publish DWH artifacts atomically

harness/tht/cli/preprocess_cmd.py

@@ -19,18 +19,29 @@ def run_dwh_from_config(
     config: Path, *, steps: tuple[str, ...], resume: str | None = None,
 ):
     from tht.cli.lsh_cmd import build_lsh_artifacts
-    from tht.cli.schema_cmd import _load_config_or_exit, refresh_catalog
-    from tht.jobs.dwh_pipeline import DwhPreprocessPipeline, fingerprint
+    from tht.cli.schema_cmd import _load_config_or_exit, physical_path, refresh_catalog
+    from tht.jobs.dwh_pipeline import DwhPreprocessPipeline, active_generation_dir, fingerprint
 
     cfg = _load_config_or_exit(config)
     workspace_root = cfg.paths.artifacts.parent
+    active = active_generation_dir(workspace_root)
+    lsh_names = (
+        f"{cfg.database.db_schema}_lsh.pkl",
+        f"{cfg.database.db_schema}_minhashes.pkl",
+        f"{cfg.database.db_schema}_meta.json",
+    )
     pipeline = DwhPreprocessPipeline(
         workspace_id=config.stem.lower().replace(".", "-").replace("_", "-"),
         workspace_root=workspace_root,
         config_fingerprint=fingerprint(cfg.model_dump_json()),
         input_fingerprint=fingerprint(config.resolve().as_posix()),
-        introspect=lambda: refresh_catalog(cfg),
-        build_lsh=lambda: build_lsh_artifacts(cfg),
+        introspect=lambda output: refresh_catalog(cfg, output_path=output),
+        build_lsh=lambda physical, output: build_lsh_artifacts(
+            cfg, physical_file=physical, output_dir=output
+        ),
+        lsh_filenames=lsh_names,
+        current_physical=physical_path(cfg),
+        current_lsh_dir=active if active is not None else cfg.paths.indexes / "lsh",
     )
     return pipeline.run(steps, resume_run_id=resume)