migrate to using completions endpoint by default

[cpacker]

Closes #595

---

Please describe the purpose of this pull request

Use completions endpoint for LM Studio instead of monkeypatch on chat/completions endpoint.

• Make lmstudio-legacy backend type, make default use completions
• Add note in docs about how if you're on <= 0.2.8, use lmstudio-legacy

How can we test your PR during review?

• On new configure, test lmstudio, make sure it pipes to completions
• On new configure, test lmstudio-legacy, make sure it pipes to chat/completions

Have you tested this PR?

Yes, see ^

---

LM Studio trace on completions endpoint:

 \"message\": \"More human than human is our motto.\"\n  }\n}\nFUNCTION RETURN: {\"status\": \"OK\", \"message\": null, \"time\": \"2023-12-15 12:14:09 PM \"}\nUSER: {\"type\": \"login\", \"last_login\": \"Never (first login)\", \"time\": \"2023-12-15 12:14:09 PM \"}\n### RESPONSE\nASSISTANT:\n{\n  \"function\":"
}
[2023-12-15 12:14:09.810] [INFO] Provided inference configuration: {
  "n_threads": 4,
  "n_predict": 8192,
  "top_k": 40,
  "top_p": 0.95,
  "temp": 0.8,
  "repeat_penalty": 1.1,
  "input_prefix": "<|im_end|>\n<|im_start|>user\n",
  "input_suffix": "<|im_end|>\n<|im_start|>assistant\n",
  "antiprompt": [
    "<|im_start|>",
    "<|im_end|>",
    "\nUSER:",
    "\nASSISTANT:",
    "\nFUNCTION RETURN:",
    "\nUSER",
    "\nASSISTANT",
    "\nFUNCTION RETURN",
    "\nFUNCTION",
    "\nFUNC",
    "<|im_sep|>"
  ],
  "pre_prompt": "",
  "seed": -1,
  "tfs_z": 1,
  "typical_p": 1,
  "repeat_last_n": 64,
  "frequency_penalty": 0,
  "presence_penalty": 0,
  "n_keep": 0,
  "logit_bias": {},
  "mirostat": 0,
  "mirostat_tau": 5,
  "mirostat_eta": 0.1,
  "memory_f16": true,
  "multiline_input": false,
  "penalize_nl": true
}
[2023-12-15 12:14:25.162] [INFO] Accumulated 62 tokens:  "send_message",
  "params": {
    "inner_thoughts": "I have been activated. I am now engaging with the user for the first time.",
    "message": "Hello Chad! It's great to meet you. What brings you here today?"
  }
}
[2023-12-15 12:14:25.285] [INFO] Generated prediction: {
  "id": "cmpl-",
  "object": "text_completion",
  "created": 1702671249,
  "model": "/lmstudio_models/TheBloke/OpenHermes-2.5-Mistral-7B-16k-GGUF/openhermes-2.5-mistral-7b-16k.Q8_0.gguf",
  "choices": [
    {
      "index": 0,
      "text": " \"send_message\",\n  \"params\": {\n    \"inner_thoughts\": \"I have been activated. I am now engaging with the user for the first time.\",\n    \"message\": \"Hello Chad! It's great to meet you. What brings you here today?\"\n  }\n}",
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 2873,
    "completion_tokens": 62,
    "total_tokens": 2935
  }
}

LM Studio trace on chat/completions endpoint:

[2023-12-15 12:26:48.756] [INFO] [LM STUDIO SERVER] Last message: { role: 'user', content: 'You are MemGPT, the latest version of Limnal Corporation's digital companion, developed in 2023.
You... (truncated in these logs)' } (total messages = 1)
[2023-12-15 12:27:00.315] [INFO] [LM STUDIO SERVER] Accumulating tokens ... (stream = false)
...
[2023-12-15 12:27:02.827] [INFO] [LM STUDIO SERVER] Generated prediction: {
  "id": "chatcmpl-",
  "object": "chat.completion",
  "created": 1702672008,
  "model": "/lmstudio_models/TheBloke/OpenHermes-2.5-Mistral-7B-16k-GGUF/openhermes-2.5-mistral-7b-16k.Q8_0.gguf",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "\"pause_heartbeats\",\n{\n  \"inner_thoughts\": \"Need some time to think.\",\n  \"minutes\": 5\n}\n### INPUT"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 2908,
    "completion_tokens": 34,
    "total_tokens": 2942
  }
}