feat: add seq cursor, work_items, provider fallback, and delivery reliability

Major reliability improvements to the message processing pipeline:

- Add messages.seq monotonic cursor replacing timestamp-based cursors,
  preventing message loss from timestamp collisions with LIMIT queries
- Add work_items table separating agent production from delivery,
  enabling delivery retry without re-running the agent
- Propagate Discord send failures instead of silently swallowing them
- Add Claude 429 → Kimi K2.5 automatic provider fallback with cooldown
- Fix follow-up turn state reset in live index.ts path (not just
  message-runtime.ts) to prevent final output from being lost
- Add restart context tracking for graceful restart announcements
- Lazy migration from timestamp cursors to seq cursors for existing data

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/group-queue.test.ts

@@ -414,6 +414,30 @@ describe('GroupQueue', () => {
     await vi.advanceTimersByTimeAsync(10);
   });
 
+  it('sendMessage touches active run activity after piping follow-up', async () => {
+    let resolveProcess: () => void;
+    const touch = vi.fn();
+
+    const processMessages = vi.fn(async () => {
+      await new Promise<void>((resolve) => {
+        resolveProcess = resolve;
+      });
+      return true;
+    });
+
+    queue.setProcessMessagesFn(processMessages);
+    queue.enqueueMessageCheck('group1@g.us');
+    await vi.advanceTimersByTimeAsync(10);
+    queue.registerProcess('group1@g.us', {} as any, 'agent-1', 'test-group');
+    queue.setActivityTouch('group1@g.us', touch);
+
+    expect(queue.sendMessage('group1@g.us', 'hello')).toBe(true);
+    expect(touch).toHaveBeenCalledTimes(1);
+
+    resolveProcess!();
+    await vi.advanceTimersByTimeAsync(10);
+  });
+
   it('sendMessage returns false for task agents so user messages queue up', async () => {
     let resolveTask: () => void;