fix(cache): enhance sequence generation atomicity and concurrency safety

Implement Lua scripts for atomic sequence incrementing and cold-start initialization in Redis to prevent race conditions. Improve the sequence buffer manager with CAS-like logic to prevent interval leakage during concurrent allocations. Update the message repository to use `SELECT ... FOR UPDATE` and conditional updates for `last_seq` to ensure database consistency during high concurrency.

- Add `nextSeqLua` and `initSeqLua` for atomic Redis operations
- Implement thread-safe buffer management in `seq_buffer.go`
- Add row-level locking in `message_repo.go`
- Update service layers to support group-aware sequence retrieval

internal/cache/seq_buffer.go

@@ -85,6 +85,8 @@ end
 -- 缓冲区耗尽，扩展 LAST
 local newLast = last + bufSize
 redis.call('HSET', bufKey, 'LAST', newLast)
+-- 推进 CURR 防止下一个调用者获得相同的 start seq
+redis.call('HSET', bufKey, 'CURR', curr + 1)
 -- 同步更新 msg_seq（兼容旧 INCR key，确保 READ 的值不回退）
 local oldSeq = tonumber(redis.call('GET', seqKey))
 if oldSeq and oldSeq > newLast then
@@ -178,12 +180,29 @@ func (m *SeqBufferManager) GetNextSeq(ctx context.Context, convID string, isGrou
 		return 0, fmt.Errorf("allocate from redis: %w", err)
 	}
 
-	// 3. 更新本地缓冲区
+	// 3. 更新本地缓冲区（CAS 防止并发分配时丢失区间）
 	newBuf := &localSeqBuffer{curr: startSeq, last: endSeq}
-	m.buffers.Store(convID, newBuf)
-
-	// startSeq 已经被 Redis 分配，返回它
-	return startSeq, nil
+	for {
+		existing, loaded := m.buffers.LoadOrStore(convID, newBuf)
+		if !loaded {
+			// 首次插入，直接使用
+			return startSeq, nil
+		}
+		// 已有缓冲区，检查是否被其他 goroutine 刷新
+		oldBuf := existing.(*localSeqBuffer)
+		oldBuf.mu.Lock()
+		if oldBuf.curr < oldBuf.last {
+			// 旧缓冲区仍有余量，直接使用（丢弃本次分配，避免区间泄漏）
+			seq := oldBuf.curr + 1
+			oldBuf.curr = seq
+			oldBuf.mu.Unlock()
+			return seq, nil
+		}
+		// 旧缓冲区耗尽，替换为新分配的区间
+		m.buffers.Store(convID, newBuf)
+		oldBuf.mu.Unlock()
+		return startSeq, nil
+	}
 }
 
 // allocateFromRedis 向 Redis 预分配 seq 区间