Employ the actual llm to perform the prompt token counting

[githubjacky]

The motivation is that the get_num_tokens method of the LangChain chat models is buggy, which will have this warning:

Token indices sequence length is longer than the specified maximum sequence length for this model (8599 > 1024). Running this sequence through the model will result in indexing errors

It's better to calculate the token by ourself, or maybe we can check the source code.