Expose context window for each model

[krrishdholakia]

expose the max tokens each model takes - this enables people to adjust around each model's context window

[krrishdholakia]

this is now done

[krrishdholakia]

get_max_tokens()