Add IsVariantAnd derive

[ChrisJr404]

Closes #365.

This adds an IsVariantAnd derive: for each enum variant foo it generates an is_foo_and(&self, f) -> bool method that returns true only when the value is that variant and the given closure returns true for its contents. It's the Option::is_some_and pattern generalised to any enum, which is handy for matching on a variant and checking something about its payload in a single expression instead of a matches! or a match arm.

I modelled the API on the existing IsVariant and Unwrap derives:

• The method takes &self (like IsVariant), so it's non-consuming and works on borrowed enums.
• The closure receives the variant's fields by reference as a tuple, in declaration order, the same way Unwrap returns them. A single-field variant collapses to a plain reference, so Just(T) gives is_just_and(|x: &T| ...), exactly mirroring Option::is_some_and.
• Field-less variants (unit, empty tuple () or empty record {}) take a FnOnce() -> bool closure instead.
• #[is_variant_and(ignore)] skips a variant, consistent with #[is_variant(ignore)].

It's behind a new is_variant_and feature flag (added to full), with docs in impl/doc/is_variant_and.md, a CHANGELOG entry, and integration tests in tests/is_variant_and.rs covering tuple/struct/unit variants, references, generics with bounds, ignore, a single-variant enum, and the nightly/deprecated cases the IsVariant tests exercise.

One thing worth a look during review: I went with references-in-a-tuple for the closure argument rather than consuming self like Option::is_some_and does, since a consuming derive felt surprising here and IsVariant already established &self. Happy to change the shape if you'd prefer something else.

[ChrisJr404]

The failing stable check was the eq/non_eq_field compile_fail snapshot, not this derive. Newer stable rustc dropped the 'see issue #48214' help line, so I reblessed that .stderr to match. Passes locally now with --features full.