Refactor inference engine

[kyle65463]

Overview

• Support passing kwargs to inference engine to make its usage more flexible
• Move system_prompt from vectorQ config to initialization of inference engines
• Refine comments of vectorQ policies and inference engines
• Rename open_ai.py and lang_chain.py to openai.py and langchain.py to follow their naming convention
• Rename create method of InferenceEngine to infer to follow the naming of VectorQ class
• Rewrote tests for inference engines and add tests for using kwargs

Passing **kwargs for inference engine

With the new interface, the user can pass additional parameters to the underlying inference engine to get more flexibility, for example:

vectorq.infer(
   prompt="What is the capital of France?",
   response_format={...}, # an argument of OpenAI's API
)

Or overriding the model/temperature of the inference engine initialization:

inference_engine = OpenAIInferenceEngine(
  model_name="gpt-4.1-nano",
  temperature=0.0,
)
...
vectorq.infer(
  prompt="What is the capital of France?",
  model_name="gpt-4o", # override when doing inference
  temperature=1.0, # override when doing inference
)