Replace huggingface-cli download command with simple https client to pull models

[swarajpande5]

Resolves #335

This PR replaces huggingface-cli with the same download_file() function as used in ollama.py. This makes the pull output similar for huggingface:// and ollama:// transports.
However, the depdency on huggingface-cli is not fully removed. If huggingface-cli is installed then it can be used for other commands (such as login, logout, push) but not for pulling the models. If huggingface-cli is not present, then a simple message will be printed on the console for those other commands.

Example run:

(venv) spande-mac:ramalama (hf-https-pulls*) $ ./bin/ramalama pull huggingface://TheBloke/Tinyllama-2-1b-miniguanaco-GGUF/tinyllama-2-1b-miniguanaco.Q2_K.gguf
Pulling sha256:17b5d82a52b71b74ce303fbe17fbba66462d80fb8b42b80629012b58c1a0b6a7: 100% ▕####################▏ 460M/460M 3.30MB/s 00:00 

(venv) spande-mac:ramalama (hf-https-pulls*) $ tree $HOME/.local/share/ramalama    
/Users/spande/.local/share/ramalama
├── models
│   ├── huggingface
│   │   └── TheBloke
│   │       └── Tinyllama-2-1b-miniguanaco-GGUF
│   │           └── tinyllama-2-1b-miniguanaco.Q2_K.gguf -> ../../../../repos/huggingface/TheBloke/Tinyllama-2-1b-miniguanaco-GGUF/tinyllama-2-1b-miniguanaco.Q2_K.gguf/sha256:17b5d82a52b71b74ce303fbe17fbba66462d80fb8b42b80629012b58c1a0b6a7
│   ├── oci
│   └── ollama
└── repos
    ├── huggingface
    │   └── TheBloke
    │       └── Tinyllama-2-1b-miniguanaco-GGUF
    │           └── tinyllama-2-1b-miniguanaco.Q2_K.gguf
    │               └── sha256:17b5d82a52b71b74ce303fbe17fbba66462d80fb8b42b80629012b58c1a0b6a7
    ├── oci
    └── ollama

14 directories, 2 files