Move to node-unrar-js for cbr and node-stream-zip for cbz

[mikiher]

This is a continuation of PR #3422, and more comprehensively fixes #3406, fixes #3399, and fixes #2526.

I looked for a replacement for libarchive for extraction from cbr (rar) files. I found node-unrar-js which seems to work similarly to libarchive (with the extraction algorithm implemented in a wasm module), and which seems to work well on a few files that libarchive failed on, and additionally doesn't require loading the full cbr into a buffer. The npm package is minimal and has 0 dependencies.

Additionally, I moved from libarchive to our existing libs/nodeStreamZip implementation for cbz files.

The old libarchive implementation as well as the new implementations are all wrapped with the same interface, and parseComicMetadata was modified to use that interface.

I'm temporarily keeping the libarchive implementation so that we can move back to it in case we have any serious issues with the new implementations. Will remove it in a couple of weeks.

[Nda70]

Thanks a lot for your hard work and perseverance.

[mikiher]

The last commit cleans up empty directories in the extraction temp dir (this can happen when the extracted file is not in the archive's root)

[advplyr]

I tested on my comic library and it is working well.

Since node-unrar-js is using WASM it could also be used in the client, while libarchive is still used for cbz's.
The comic reader could be refactored to get each page/image from the server. The only issue with that is handling downloaded comics in the mobile apps.

Thanks!