feat: Semantische Suche über Chat-History (memory_search)

- Lokales Embedding-Model (all-MiniLM-L6-v2, ONNX, CPU)
- EmbeddingService: WordPiece-Tokenizer + Mean Pooling (384 Dim)
- VectorSearchService: SQLite-basierte Vektor-Speicherung + Cosine Similarity
- EmbeddingBackgroundService: Backfill bestehender Messages + Live-Queue
- Neues Tool memory_search: Semantische Suche über alle Sessions (inkl. archivierte)
- ~9000 Messages embedded, Suche über komplette Chat-History

Endpoints/ChatEndpoint.cs

@@ -22,7 +22,7 @@ public static class ChatEndpoint
         app.MapPost("/api/chat/stop", HandleStop);
     }
 
-    private static async Task HandleChat(HttpRequest req, HttpResponse res, IConfiguration config, ChatService chatService, DatabaseService db, StreamingTracker streamingTracker, AnthropicService anthropic, RemoteConnectionManager remoteMgr, CancellationToken cancellationToken)
+    private static async Task HandleChat(HttpRequest req, HttpResponse res, IConfiguration config, ChatService chatService, DatabaseService db, StreamingTracker streamingTracker, AnthropicService anthropic, RemoteConnectionManager remoteMgr, EmbeddingBackgroundService embeddingBg, CancellationToken cancellationToken)
     {
         var body = await JsonSerializer.DeserializeAsync<ChatRequest>(req.Body);
         if (body == null)
@@ -108,7 +108,7 @@ public static class ChatEndpoint
                 Data = b.Data
             }).ToList();
 
-            await db.SaveMessageWithAttachments(new MessageRecord
+            var userMsgId = await db.SaveMessageWithAttachments(new MessageRecord
             {
                 ChainId = ctx.ChainId,
                 Iteration = 0,
@@ -116,6 +116,7 @@ public static class ChatEndpoint
                 Type = "text",
                 Content = textContent
             }, attachmentRecords);
+            embeddingBg?.Enqueue(userMsgId, ctx.ChainId, ctx.SessionId, ctx.AgentId, "user", textContent ?? "");
         }
 
         res.Headers["Content-Type"] = "text/event-stream";