Find a more robust way to handle the tar extraction of the downloaded files

[PaulRBerg]

The current way the repo files are extracted after they are downloaded is prone to errors. I've spent four hours today debugging https://github.com/paulrberg/create-eth-app/issues/135, thinking that the issue is related to Jest or GitHub Actions, whereas in fact that was caused by a bug in the getRefs function. Read the comments I left in the issue for more details.

To-dos:

• [ ] Find or build an external package that can handle the parsing of the GitHub ref used to download the repo files from codeload.github.com
• [ ] Make the getRefs function less bloated

[PaulRBerg]

See Node.js package for extracting branch name from git ref.

[PaulRBerg]

The getRefs must definitely be overhauled such that ref is always the branch name itself, and it is not prefixed with refs/heads. Ditto for tags, remotes, etc.

[Adsgw]