feat: add /models/apply endpoint to prepare models

[mudler]

This change allow to use the API to programmatically download models from a "gallery'.
It doesn't introduce the concept of the gallery, however, it introduces a way to install models from pre-defined yaml files which would be part of it, so it provides the "base" building block.

It exposes an api endpoint to apply new models from gallery YAML files. gallery files includes information to retrieve the model, templates to use, and its configuration file.

For instance, consider:

name: "groovy"
description: |
    LocalAI GPT4ALL-J Model
license: "Apache 2.0"
urls:
- gpt4all.io

config_file: |
    parameters:
      model: groovy
      top_k: 80
      temperature: 0.2
      top_p: 0.7
    context_size: 1024
    stopwords:
    - "HUMAN:"
    - "GPT:"
    roles:
      user: ""
      system: ""
    template:
      completion: "groovy-completion"
      chat: groovy-chat

files:
    - filename: "groovy"
      sha: ""
      uri: "https://gpt4all.io/models/ggml-gpt4all-j-v1.3-groovy.bin"

prompt_templates:
    - name: "groovy-completion"
      content: |
        Complete the prompt
        ### Prompt:
        {{.Input}}
        ### Response:
    - name: "groovy-chat"
      content: |
        The prompt below is a question to answer, a task to complete, or a conversation to respond to; decide which and write an appropriate response.
        ### Prompt:
        {{.Input}}
        ### Response:

uploaded as a gist, could be used as so to configure LocalAI to download all the necessary files in runtime to start using the model:

$ curl http://localhost:8080/models/apply -H "Content-Type: application/json" -d '{
     "url": "https://gist.githubusercontent.com/mudler/6112666e77061fe35ca3a535d25ccddc/raw/45eabb442f25db5bc991f2d2a616c15d4c26be67/gpt4all-j-localai.yaml"
   }'
{"uid":"1059474d-f4f9-11ed-8d99-c4cbe106d571","status":"http://localhost:8080/models/jobs/1059474d-f4f9-11ed-8d99-c4cbe106d571"}
$ curl http://localhost:8080/models/jobs/1059474d-f4f9-11ed-8d99-c4cbe106d571      
{"error":null,"processed":true,"message":"completed"}

This will:

• Start a batch job to download the model and the relevant files
• Reload the models once finished to pick up the model

In a bash script, you can wait on the operation to finish by checking the processed field:

model_url="https://gist.githubusercontent.com/mudler/6112666e77061fe35ca3a535d25ccddc/raw/45eabb442f25db5bc991f2d2a616c15d4c26be67/gpt4all-j-localai.yaml"

response=$(curl -s http://localhost:8080/models/apply -H "Content-Type: application/json" -d '{"url": "$model_url"}')

job_id=$(echo "$response" | jq -r '.uid')

while [ "$(curl -s http://localhost:8080/models/jobs/"$job_id" | jq -r '.processed')" != "true" ]; do 
  sleep 1
done

echo "Job completed"

Related to: #100