Caching of file extraction step

[ronaldtse]

On every run, a tebako-compiled executable extracts the DwarFS image out into a temporary directory for execution.

We should cache this extraction step across runs, so we only extract the files when we need to:

• e.g. a file is gone/corrupt, or the DwarFS image changed

[maxirmx]

tebako-compiled executable DOES NOT EXTRACT the DwarFS image out into a temporary directory for execution
It extracts some files that are directly accessed by Ruby extensions as expained here: #42 (comment)

If we want just extract the image we do not need DwarFS. Now we are following ruby-packer approcach and run the application from in-memory DwarFS filesystem.
Otherwise we could just compress ruby folder and extract it on startup just as OCRA does.

[ronaldtse]

This issue applies to the part that is extracted, not the part that is accessible via DwarFS:

It extracts some files that are directly accessed by Ruby extensions

For the extracted part, we want the ability to cache/persist across calls if possible (eg at a persistent cache location).