feat: install local engine

[vansangpfiev]

Describe Your Changes

Syntax:
-s, --source: local folder source to install engine

cortex engines install <engine_name> -s <source_path>

• If source_path contains engine → unzip engine to installation directory
• If source_path does not contain engine → fallback to download engine path

The same logic for CUDA dependencies

Fixes Issues

• epic: llama.cpp is installed by default #1217

Self Checklist

• Added relevant comments, esp in complex areas
• Updated docs (for bug fixes / features)
• Created issues for follow-up changes or refactoring needed

[namchuai]

✅ some small nits. Looks good to me overally!

[dan-menlo]

Lgtm, thanks @vansangpfiev. A few follow-up questions:

• For --source, will it take a folder or a file?
• Is it a folder since CUDA dependencies and llamacpp tar.gz are different files?

I will let the engineers scrutinize for code convention.

[vansangpfiev]

Lgtm, thanks @vansangpfiev. A few follow-up questions:

• For --source, will it take a folder or a file?
• Is it a folder since CUDA dependencies and llamacpp tar.gz are different files?

I will let the engineers scrutinize for code convention.

It is a folder