Fix token usage with jump forward

[comaniac]

close #173

This PR fixes the incorrect token usage when jump forward is enabled. Specifically, we introduce a new field orig_prompt_tokens, which will be set when the first jump forward happens so that we could know the original number of prompt tokens. When returning a response (a chunk in streaming or a complete response), we use the following equations to correct the token usage:

completion_tokens = curr_prompt_token - orig_prompt_tokens + completion_tokens
prompt_tokens = orig_prompt_tokens

[hnyls2002]

@comaniac Hi, I suggest initializing the orig_prompt_tokens when constructing the Req so that we can simplify the code.

[comaniac]

Thanks that makes sense. Meanwhile, can you add some comments back to explain why we need to calculate the token usage in this way?