Implement Node Generator Module for vllm Serving API Integration

[korjsh]

Purpose of Development

• When using the existing vllm module, each test requires model reinitialization, leading to significant test time overhead.
• By enabling API-based connections, similar to the OpenAI module, model reinitialization time can be saved, improving testing efficiency.

---

Testing Instructions

1. Start the vllm serving server:

   vllm serve Qwen/Qwen2.5-14B-Instruct-AWQ -q awq --port 8012
   

2. Here is sample of evaluate_config.yaml file.

       - node_type: generator
         strategy:
            metrics: 
                - metric_name: rouge
          modules:
                - module_type: vllm_api
                  uri: http://localhost:8012
                  llm: Qwen/Qwen2.5-14B-Instruct-AWQ
                  temperature: [0, 0.5]
                  max_tokens: 400

3. Execute the evaluation process.

[vkehfdl1]

@korjsh Hello! Thanks for the contribution.
Before merging, we have to make a two more things.

1. The test code. You should make a test code at the tests/autorag/nodes/generator
2. The docs. You should make a new documentation at docs/source/nodes/generator

If you feel overwhelmed to make the test code and docs, we can cover it.
If you want the PR merge faster, it is always great to do it yourself.

Thanks a lot :)