feat: enable streaming responses like OpenClaw

- Add sendStreamingMessage() to message-sender.js with typing indicators
- Enable stream: true in chatWithAI() with SSE parsing
- Replace all ctx.reply() calls with sendStreamingMessage()
- Real-time text streaming with 50ms delay between chunks

src/bot/message-sender.js

@@ -42,7 +42,42 @@ export async function sendFormatted(ctx, text) {
   }
 }
 
-export async function sendLongMessage(ctx, text) {
+export async function sendStreamingMessage(ctx, text, options = {}) {
   if (!text) return;
-  return sendFormatted(ctx, text);
+
+  const { delay = 50, maxChunkSize = 1000 } = options;
+
+  try {
+    // Stream chunks with typing indicator
+    const chunks = splitMessage(text);
+    let delayTimer = null;
+
+    for (const chunk of chunks) {
+      // Clear previous typing indicator
+      if (delayTimer) {
+        clearTimeout(delayTimer);
+        delayTimer = null;
+      }
+
+      // Send chunk with minimal delay for real-time effect
+      await ctx.reply(chunk, { parse_mode: 'Markdown' });
+
+      // Show typing indicator between chunks
+      if (chunks.length > 1) {
+        delayTimer = setTimeout(() => {
+          ctx.api.sendChatAction(ctx.chat.id, 'typing').catch(() => {});
+        }, delay);
+      }
+    }
+
+    // Clear typing indicator
+    if (delayTimer) {
+      clearTimeout(delayTimer);
+      delayTimer = null;
+    }
+  } catch (error) {
+    logger.error('Streaming send failed:', error);
+    // Fallback to non-streaming
+    await sendFormatted(ctx, text);
+  }
 }