Extends lm.inspect_history to print usage and reasoning content

[nguyenhoan1988]

Enhanced the pretty_print_history function in lm.inspect_history to include:

Usage Details: Prints response usage information when available
Reasoning Content: Displays reasoning content from the response message if present.

[arnavsinghvi11]

Hi @nguyenhoan1988 , thanks for the PR! could this instead be supported like how we do image & audio in inspect_history rather than keeping a separate verbose flag? Also, we don't need to introduce new colors - all outputs should be green

[chenmoneygithub]

Thanks for the PR!

I am not quite sure if usage on console print is useful, programmatically we do provide the Prediction.get_lm_usage() to get lm usage for the module call, and for a breakdown with good visual, we recommend using MLflow tracing.

Reasoning print is useful.