v0.7.0

Performance

• Major rewrite of code base results in better single threaded performance.
• Added rayon for even better performance.

=> 30% faster performance for generating package list with licenses.

I am considering bringing back the cache implementation from prior versions.

Let me tell you something: The implementation with smol was harder and less performant than the relatively bad blocking implementation from before. Why? I don't knoooooow. help

Edit (2025-05-06): The reason is, that said step only takes up 8ms of 310ms. That is why using rayon also did not make much sense.
In hindsight, I am not even sure why said rewrite made it so much faster.

Changed

• Packages are now sorted by their name and version.