Add debug info to help us understand the size in tokens of our prompts and responses

[jwmatthews]

It would be helpful if we had a means of enabling debug information to share how many tokens we are using up for input prompts and for output.

For example, we notice some models are stating they have a max of 1536 output tokens, we are seeing some cases of outputs that stop and we are not sure if we hit a limit on output tokens or if something else is happening.

[shawn-hurley]

I think this is a duplicate of #373 do you think that we can close in favor of that issue or is this something else?

[fabianvf]

closing in favor of #373