docs: do not use tokenizer in client for llm streaming example (#6013)

docs: do not use tokenizer in client for llm streaming example (#6013) #5749