[NFC] Reduce kwarg duplication

[christiangnrd]

This is essentially #38 for the rest of the package, as well as some opinionated (but non-functional) related changes.

Pros:

• Reduce duplication of defaults arguments, potentially reducing future typo-related bugs.
• Cleaner code for those familiar with Julia

Cons:

• More coupling of code (although different defaults can easily be added where needed
• Potentially harder to read for those unfamiliar with the code base.
• I don't think it does, but I have not tested if this has any impact on performance

@anicusan I wrote this up to be helpful, none of this is necessary for my GSOC objectives. If you like it, great! If not, no problem, I'll just close this PR. I do think bringing down the "sources of truth" could help with preventing bugs from copy-paste or things like #37.

[anicusan]

My only problem is that the kwargs... will silently accept typos, e.g. AK.foreachindex(..., max_threads=1) instead of max_tasks, then silently ignore unnecessary kwargs. It is a bit of duplication on our dev side, but it avoids unexpected behaviours for users - and shows useful errors in case of typos / wrong kwargs.

But internally - forwarding kwargs from our exposed functions to the internal implementations - I think this is a good idea, if you'd be okay with updating this PR (including the new multithreaded algorithms in 0.4.0 - sorry 😅)

[christiangnrd]

It should error once it reaches the implementation. Maybe I can add a test for every public method to ensure that it properly throws an ArgumentError?

I'll rebase #38 since this PR depends on it, and once that's merged I'll take care of this one

[christiangnrd]

@anicusan This is ready for final review. I probably shouldn't have force-pushed so I'll detail the changes I've made.

Changes mixed in with the force push:

• Rebased on 0.4
• Added tests for undefined kwargs

And then I also applied some of the same refactors into the new 0.4 files, and I removed _reduce_impl in favour of directly calling _mapreduce_impl(identity, ...) since after the kwarg changes it felt unecessary.

Please let me know before merging, I want to clean up the commit history beforehand.

[christiangnrd]

oneAPI error seems like an upstream issue. I've encountered it before

[anicusan]

This is great, thanks for also adding the MethodError tests. It's good to merge from my side - feel free to do any commit cleanup you'd like :)

[christiangnrd]

Merge away!

[anicusan]

Really appreciate this, @christiangnrd