feat: Download model simultaneously via API

[louis-jan]

Problem Statement

It would be great to allow users to download models simultaneously via the API. Currently, a POST request to /models/pull triggers only one download at a time, causing the second request to wait until the first download is completed. This causes a poor UX when integrating cortex.cpp into Jan.

Feature Idea

Supporting that might lead to a slight change in the response structure of the /events

{
  "task": {
    "id": "tinyllama:gguf",
    "items": [
      {
        "bytes": 668788096,
        "checksum": "N/A",
        "downloadUrl": "https://huggingface.co/cortexso/tinyllama/resolve/gguf/model.gguf",
        "downloadedBytes": 0,
        "id": "model.gguf",
        "localPath": "/home/jan/cortexcpp/models/cortex.so/tinyllama/gguf/model.gguf"
      },
      {
        "bytes": 668788096,
        "checksum": "N/A",
        "downloadUrl": "https://huggingface.co/cortexso/tinyllama/resolve/gguf/model.yml",
        "downloadedBytes": 651208498,
        "id": "model.yml",
        "localPath": "/home/jan/cortexcpp/models/cortex.so/tinyllama/gguf/model.yml"
      }
    ],
    "type": "Model"
  },
  "type": "DownloadUpdated"
}

So that there should be an array of tasks instead of a task object. This could affect Jan later, so we should update accordingly.

[gabrielle-ong]

Thanks @namchuai and @louis-jan! Fixed on Jan 0.5.9 simultaneous download models