Add params of config in create

[nsorros]

Description

This PR allows you to easily deploy models in huggingface inference endpoints by modifying some default params instead of having to construct a valid config json. Among other things we can now deploy an LLM as follows

hugie endpoint create \
    --repository tiiuae/falcon-7b \
    --name falcon-7b \
    --instance-type g5.2xlarge \
    --instance-size medium \
    --accelerator gpu \
    --image tgi

Fixes #74

Checklist

• Tests added
• Relevant issue mentioned
• Documentation updated

[ivyleavedtoadflax]

LGTM. Don't forget the update version number and push to pypi