Implement cryptographic hashing via various algorithms

[CompeyDev]

What's in this PR?

Closes #105.

Refer to the above issue for tracking/implementation discussions.

TODO

• Implement a base structure/framework
  • Macros for less boilerplate
  • Trait implementations
    • LuaUserData
    • FromLua
    • From for convenience
    • CryptoAlgo enum
• Implement specific algorithms
  • Md5
  • Sha(1, 256, 512, 3_256, 3_512)
  • Blake(2, 3)
  • Argon2id
  • PBKDF2
• Tests & QA
  • Agree on where to have the API be available and in what form
  • Tests
  • Documentation

[CompeyDev]

@RustCrypto doesn't have a Blake3 implementation for Rust, so it might be a bit tricky to interop with our current framework being used for this.

[filiptibell]

Thank you for the PR! However, I'm not really sure if this is what I'd like hashing/crypto algorithms to look like in Lune.

From a users perspective:

• A user thinking "I want to hash something" and then needing to keep track of a hasher object and learn what a digest is etc seems out of place for Lune - we tend to restrict API surface somewhat over going for maximum flexibility, to try to not expose too much to the user and not add cognitive overhead
• Generally, to represent enum-like options in our built-in libraries, we use string literal unions as arguments (serde.encode/decode, stdio.prompt, ...), and not standalone functions. Naming functions this way prioritizes what the function does instead of how it does it
• We don't have this kind of library-inside-library (crypto inside serde) anywhere else

I think this still needs some additional thought and design before moving on. The implementation using macros, boxing, mutexes seems more complex than it needs to be. Luau now also has a native buffer type, which negates some of the performance concerns we previously had for an API more similar to serde.encode/decode. I'll be leaving a comment on the related issue as well and we can try to keep design discussion there