Report dollars spent on OpenAI

[kierangilliam]

The user does not have visibility into the cost of each session. This PR adds a report_metrics callback for environments to display useful metrics to the user.

This has been implemented for the TaskSolving environment. Try "brainstorming" to see the result.

$ python main_tasksolving_cli.py --task tasksolving/brainstorming
...
Agent spend:  
Agent (Role: AGENT_TYPES.ROLE_ASSIGNMENT) role assigner: $0.001631  
Agent (Role: AGENT_TYPES.SOLVER) A mechanical engineer specialized in hydrogen storage technologies: $0.01  
Agent (Role: AGENT_TYPES.CRITIC) An electrical engineer with expertise in renewable energy integration: $0.002690  
Agent (Role: AGENT_TYPES.CRITIC) A regulatory specialist familiar with hydrogen storage regulations: $0.002037  
Agent (Role: AGENT_TYPES.CRITIC) An infrastructure planner with experience in large-scale projects: $0.002518  
Agent (Role: AGENT_TYPES.EXECUTION) Dummy Executor: $0.000000  
Agent (Role: AGENT_TYPES.EVALUATION) Evaluator: $0.003592  
Total spent: $0.018993 

[chenweize1998]

Awesome! Thanks for your help