feat(wren-ai-service): semantics description pipeline to generate or optimize the user description

[paopa]

This PR introduces the SemanticsDescription service, which provides endpoints for generating and optimizing semantic descriptions based on user prompts and selected models.

Web API Specification

1. Generate Semantic Description

Endpoint: POST /v1/semantics-descriptions

Request Body:

{
    "selected_models": ["model1", "model2"],
    "user_prompt": "Describe the data model",
    "mdl": "..."
}

Response:

{
    "id": "unique_id",
}

2. Retrieve Semantic Description

Endpoint: GET /v1/semantics-descriptions/{id}

Path Parameter:

• id: Unique identifier of the semantic description resource

Response:

{
   "id":"unique_id",
   "status":"finished",
   "response":{
      "model1":{
         "name":"model1",
         "columns":[
            {
               "name":"col1",
               "properties":{
                  "description":"Unique identifier for each record in the example model."
               }
            }
         ],
         "properties":{
            "description":"This model is used for analysis purposes, capturing key attributes of records."
         }
      }
   }
}

or, in case of an error:

{
    "id": "unique_id",
    "status": "failed",
    "error": {
        "code": "OTHERS",
        "message": "Error description"
    }
}

Usage

1. Call the POST endpoint to initiate a semantic description generation.
2. Use the returned ID to poll the GET endpoint until the status is "finished" or "failed".
3. Once finished, retrieve the generated description from the "response" field.

Notes

• The generation process may take some time, so implement appropriate polling intervals when checking the status.
• Error handling has been implemented to catch and report various exceptions that may occur during the process.

Screenshots

[cyyeh]

basically LGTM, thanks for hard work.