feat: align function id with tool call response

[drbh]

This PR improve mapping of tool ids to responses and enables tool_call_ids to be included in templates

Changes:

1. maps input tool.id to tool_calls[0].id for chosen tool if tool.id is sepcified
2. includes tool_call_id in chat template in the case an id is required

1. map tool id to tool response

import requests
import json

messages = [
    {
        "role": "system",
        "content": "You're a human assistant helping a user with a task. Use the tools to complete the task.",
    },
    {"role": "user", "content": "What's the weather like in Boston today?"},
]

payload = dict(
    messages=messages,
    seed=42,
    max_tokens=40,
    tools=[
        {
            "type": "function",
            "function": {
                "id": "TOOL_ID_01",
                "name": "get_current_weather",
                "description": "Get the current weather",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "The city and state, e.g. San Francisco, CA",
                        },
                        "format": {
                            "type": "string",
                            "enum": ["celsius", "fahrenheit"],
                            "description": "The temperature unit to use. Infer this from the users location.",
                        },
                    },
                    "required": ["location", "format"],
                },
            },
        }
    ],
)

response = requests.post(
    "http://127.0.0.1:3000/v1/chat/completions",
    json=payload,
    headers={"Content-Type": "application/json"},
)

print(json.dumps(response.json(), indent=2))

returns a response with the specified tool id

{
  "object": "chat.completion",
  "id": "",
  "created": 1741893608,
  "model": "mistralai/Mistral-7B-Instruct-v0.3",
  "system_fingerprint": "3.2.1-dev0-native",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "tool_calls": [
          {
            "id": "TOOL_ID_01",
            "type": "function",
            "function": {
              "description": null,
              "name": "get_current_weather",
              "arguments": "{\"location\":\"Boston, MA\",\"format\":\"celsius\"}"
            }
          }
        ]
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 261,
    "completion_tokens": 34,
    "total_tokens": 295
  }
}

2. include tool_call_id in chat template

import requests
import json

messages = [
    {
        "role": "system",
        "content": "You're a human assistant helping a user with a task. Use the tools to complete the task.",
    },
    {"role": "user", "content": "What's the weather like in Boston today?"},
    {
        "role": "assistant",
        "tool_calls": [
            {
                "id": "TOOL_ID_1",
                "type": "function",
                "function": {
                    "description": None,
                    "name": "get_current_weather",
                    "arguments": '{"location":"Boston, MA","format":"celsius"}',
                },
            }
        ],
    },
    {
        "role": "tool",
        "tool_call_id": "TOOL_ID_1",
        "content": "-1C",
    },
]

payload = dict(
    messages=messages,
    seed=42,
    max_tokens=40,
    tools=[
        {
            "type": "function",
            "function": {
                "id": "TOOL_ID_1",
                "name": "get_current_weather",
                "description": "Get the current weather",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "The city and state, e.g. San Francisco, CA",
                        },
                        "format": {
                            "type": "string",
                            "enum": ["celsius", "fahrenheit"],
                            "description": "The temperature unit to use. Infer this from the users location.",
                        },
                    },
                    "required": ["location", "format"],
                },
            },
        }
    ],
    # force a non tool response in the case the last message was a tool response
    tool_choice="none" if messages[-1]["role"] == "tool" else "auto",
)

response = requests.post(
    "http://127.0.0.1:3000/v1/chat/completions",
    json=payload,
    headers={"Content-Type": "application/json"},
)

print(json.dumps(response.json(), indent=2))

returns a natural language response that internal included the tool call id in the prompt

{
  "object": "chat.completion",
  "id": "",
  "created": 1741893739,
  "model": "mistralai/Mistral-7B-Instruct-v0.3",
  "system_fingerprint": "3.2.1-dev0-native",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The weather in Boston today is -1 degrees Celsius."
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 100,
    "completion_tokens": 14,
    "total_tokens": 114
  }
}

without these changes the response from mistralai/Mistral-7B-Instruct-v0.3 (which requires an ID of a specific length)

{
  "error": "Template error: syntax error: Tool call IDs should be alphanumeric strings with length 9! (in <string>:81)",
  "error_type": "template_error"
}