Use super equals and hashCode overrides for NonEmptyList

[RusticFlare]

Fixes #2807 by removing the overrides of equals and hashCode in NonEmptyList, so that we use the implementations in AbstractList.

One thing to note is that Kotest's shouldBe doesn't use the equals function for Iterable, so I've tested the result of == explicitly (and left a comment to that effect).

(This is my first PR here 👋 so let me know if I've missed anything)

[nomisRev]

Hey @RusticFlare,

This change looks a bit dangerous to me, since it breaks shouldBe from Kotest. I know many people rely on shouldBe on Kotest.

I am wondering if the fix for that ticket is rather not simply removing if (it::class != this::class) from the equals implementation. That way it would solve the issue from the ticket, but wouldn't break shouldBe if not mistaken.

[nomisRev]

Or is shouldBe incorrect for this kind of comparisons 🤔

[RusticFlare]

Hi, it doesn't break shouldBe. I just couldn't use shouldBe to write a failing test because shouldBe doesn't use equals on Iterable. The current behaviour is an obvious bug and I don't think there is any value in overriding the equals and hashCode on AbstractList.

[nomisRev]

Okay, thanks for clarifying @RusticFlare! I miss-understood.

We cannot break the binary, so instead of removing I think we should manually delegate to the super class for now. We can break the binary in 2.x.x. Then CI should turn green, it's now turning red on a binary breaking change.

API check failed for project arrow-core.

[RusticFlare]

@nomisRev I've made the suggested change and hopefully the comment about shouldBe is clearer now

[RusticFlare]

Looks like CI is green 🎉 I can make another PR to remove the methods for 2.x.x once this is merged.

[nomisRev]

Thank you for your first contribution @RusticFlare 🙏:clap::tada: Sorry for the hold-up.

[raulraja]

thanks @RusticFlare !