Support package-data for llm_judge

Nobody has claimed this yet.

It seems better to ensure that data for `llm_judge` is included in the package, even if installed via pip.

Install with pip
```
pip3 install "fschat[model_worker,llm_judge]"
```

Script for using llm_judge as module
```
python3 -m fastchat.llm_judge.gen_model_answer --model-path lmsys/vicuna-7b-v1.5 --model-id vicuna-7b-v1.5
```

Error
```
File "/usr/local/lib/python3.10/dist-packages/fastchat/llm_judge/gen_model_answer.py", line 288, in
run_eval(
File "/usr/local/lib/python3.10/dist-packages/fastchat/llm_judge/gen_model_answer.py", line 36, in run_eval
questions = load_questions(question_file, question_begin, question_end)
File "/usr/local/lib/python3.10/dist-packages/fastchat/llm_judge/common.py", line 91, in load_questions
with open(question_file, "r") as ques_file:
FileNotFoundError: [Errno 2] No such file or directory: 'data/mt_bench/question.jsonl'
```

The `data` directory does not exist.
```
/usr/local/lib/python3.10/dist-packages/fastchat/llm_judge
├── clean_judgment.py
├── common.py
├── compute_agreement.py
├── download_mt_bench_pregenerated.py
├── gen_api_answer.py
├── gen_judgment.py
├── gen_model_answer.py
├── qa_browser.py
└── show_result.py
```

Expected solution
```
# pyproject.toml
...

[tool.setuptools.package-data]
fastchat.llm_judge = ["data/*"]

...
```

Start with pyproject.toml and inspect the fastchat/llm_judge/data directory and its packaging configuration. Install the project with the llm_judge extra using pip, then run the reported python3 -m fastchat.llm_judge.gen_model_answer command. Done means the installed package contains the data files and no longer raises FileNotFoundError for data/mt_bench/question.jsonl.

Written by the indexing model from the issue text.