Context

When downloading a sharded model, files are downloaded in parallel, it was added in:

• common: llama_load_model_from_url split support  #6192

The progressions of each download conflict:

Need to properly implement CURLOPT_NOPROGRESS for parallel download.

Example in #6515:

main --hf-repo ggml-org/models \
  --hf-file grok-1/grok-1-q4_0-00001-of-00009.gguf \
  --model   models/grok-1-q4_0-00001-of-00009.gguf \
  -ngl 64
   --prompt "I believe the meaning of life is"