Incorrect handling of types matched as idents

[Cerber-Ursi]

The issue was originally discovered on URLO. Minimal example:

macro_rules! impl_id {
    ($typename:ident) => {
        #[juniper::graphql_object]
        impl $typename { // ERROR: could not determine ident for the `impl` type
            fn id(&self) -> i32 {
                42
            }
        }
    };
}

struct Unit;
impl_id!(Unit);

The reason for this error is the fact that, after being matched as ident, the identifier is not a bare ident anymore, it's a syn::TypeGroup, which is not handled correctly.

I've added the example above as test and modified the TypeExt::unparenthesized accordingly. It's entirely possible, however, that there are other places where either explicit match for TypeGroup or the call to unparenthesized should be added; I'm just not sure how to generate realistic test cases similar to the one above.

---

Side note: I wasn't able to run tests for this workspace, until I added features chrono and chrono-tz to the invocation. It's likely the consequence of #1010, and I'm not sure whether this is intentional (or desirable).

[tyranron]

@Cerber-Ursi thanks! 🍻

Very valuable addition. Didn't know about that nuance.