feat(is_variant): add #[must_use] annotation

[TheLostLambda]

Resolves #349

Synopsis & Solution

Adds a #[must_use] annotation to the static methods generated by IsVariant. I also refactored the derive to use matches!(...) instead of a manual match block with bools, but that's just a style tweak.

Checklist

• Documentation is updated (if required)
• Tests are added/updated (if required)

I'll be honest, I'm not 100% sure how to write tests for this one, but I have a screenshot of it in action:

• CHANGELOG entry is added (if required)

[tyranron]

@TheLostLambda thanks for quick reaction!

Let's add few bikesheddings and get it merged!

[tyranron]

For hygiene, this should be used like derive_more::core::matches!.

[tyranron]

I'll be honest, I'm not 100% sure how to write tests for this one, but I have a screenshot of it in action:

You can add tests/compile_fail with #[forbid(unused_must_use)]. So if this will be removed in future, the test case will compile successfully, and so, fail the test.

[TheLostLambda]

@tyranron Ah! Excellent! The tests/compile_fail is exactly what I was looking for!

I'll get to work addressing these loose ends!

[TheLostLambda]

I just went a little insane trying to figure out why those tests weren't running for me, but it seems the compile fail tests are all disabled on nightly Rust (only running for stable).

I think that means all of these tests are missed by CI as well, since there are a couple of failing tests unrelated to my change here.

Would you like me to add something to the CI in this PR or remove that stable-only attribute? I assume that was added in the first place to stop things from breaking when the nightly and stable compilers gave different errors, so perhaps just adding a bit in the CI is the best thing to do?

EDIT: Actually, they look like they should be running? Let me investigate...

[TheLostLambda]

Conclusion is that I have no idea why that last CI run didn't fail... It looks like things are being run in the stable test CI (one of the matrix options), but it passed in CI and failed for me, with exactly the same version of stable Rust?

I've pushed the changes that make things pass for me locally, so maybe those will break the CI version? They should?

TL;DR is that I think I've addressed all of your comments, but, unrelated, I'm not sure if the compile_fail tests are working correctly either in CI or on my machine...

[tyranron]

@TheLostLambda this is just some platform-dependent output weirdness.

Overall, now it's LGTM. Thanks!