Improve the benchmark by evaluating multiple models and display the results

[bugsz]

Closes #

📑 Description

As the title suggests:

1. Support evaluating multiple models at the same time, simply by sotopia benchmark-all --model-list gpt-4o --model-list gpt-3.5-turbo, or just go ahead with the default model names.
2. Support displaying and saving the results in format in https://github.com/sotopia-lab/sotopia-space/blob/main/data_dir/models_vs_gpt35.jsonl by sotopia benchmark-display. (Seems there is no requirement for pandas so I am not sure how to display in a structured way in CLI)

✅ Checks

• [ ] My pull request adheres to the code style of this project
• [ ] My code requires changes to the documentation
• [ ] I have updated the documentation as required
• [ ] All the tests have passed
• [ ] Branch name follows type/descript (e.g. feature/add-llm-agents)
• [ ] Ready for code review

ℹ Additional Information

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 14.58333% with 41 lines in your changes missing coverage. Please review.

Project coverage is 61.03%. Comparing base (701f2a8) to head (8fb172a).

@@            Coverage Diff             @@
##             main     #126      +/-   ##
==========================================
- Coverage   61.71%   61.03%   -0.69%     
==========================================
  Files          55       55              
  Lines        2714     2756      +42     
==========================================
+ Hits         1675     1682       +7     
- Misses       1039     1074      +35     

Files	Coverage Δ
sotopia/cli/benchmark/benchmark.py	21.17% <14.58%> (-2.27%)	:arrow_down:

... and 1 file with indirect coverage changes