fix(go-agent): finish retrieval component (#17845)

- Prefer canonical `dataset_ids` over legacy `kb_ids` in Canvas
retrieval components.
- Cover selected and explicitly cleared dataset IDs with regression
tests.
- Add cross-language retrieval with dataset-specific embedding and
rerank models.
- Support vector and keyword similarity controls, metadata filtering,
TOC enhancement, and child-chunk expansion.
- Route Canvas retrieval across datasets and memories with compatible
embedding validation.

cmd/ragflow_server.go

@@ -26,6 +26,7 @@ import (
 	"ragflow/internal/admin"
 	"ragflow/internal/agent/audio"
 	"ragflow/internal/agent/canvas"
+	"ragflow/internal/agent/retrievalbridge"
 	agenttool "ragflow/internal/agent/tool"
 	"ragflow/internal/channels"
 	"ragflow/internal/handler"
@@ -728,7 +729,14 @@ func startServer(ctx context.Context) {
 	// Initialize doc engine for skill search
 	docEngine := engine.Get()
 	documentDAO := dao.NewDocumentDAO()
-	agenttool.SetRetrievalService(agenttool.NewNLPRetrievalAdapterFromDeps(docEngine, documentDAO))
+	retrievalEnhancer := retrievalbridge.NewEnhancer(docEngine, metadataService)
+	agenttool.SetRetrievalService(agenttool.NewNLPRetrievalAdapterFromDeps(
+		docEngine,
+		documentDAO,
+		modelProviderService,
+		retrievalEnhancer,
+	))
+	agenttool.SetMemoryRetrievalService(retrievalbridge.NewMemoryAdapter(memoryService))
 	common.Info("agent: retrieval service adapter installed")
 
 	// Initialize handler layer