Feat: Allow lifetime for bounds in non-binded generic params

[stegaBOB]

It's currently not possible to use higher rank trait bounds without providing custom predicate type bounds. Our particular use case is enabling "trivial bounds" without requiring the nightly feature. For context: rust-lang/rust#48214 (comment)

For example:

// fails due to "the trait bound `std::string::String: std::marker::Copy` is not satisfied"
#[derive_where(Clone, Copy; String)]
struct TestThing {
    a: String,
}

// Compiles! Copy is just not implemented due to String obviously not being copy, so it would compile error on use.
#[derive_where(Clone, Copy; for<'a> String)]
struct TestThing {
    a: String,
}

This is useful for adding perfect derives to macro-generated structs.

I'm not fully sure whether having a HRTB should make has_type_param return false or not. I think its probably fine to keep it returning true since adding one that's not used doesn't change the behavior; it should still be derivable w/ the standard derive. Users can always make type aliases to get around this anyways if they disagree lol.

[daxpedda]

This is indeed a very strange quirk of Rust!

Please add a changelog entry as well.
Otherwise LGTM!

[daxpedda]

• Run cargo +nightly fmt.
• Changelog entry.

[stegaBOB]

Completely missed the changelog request! Sorry about that. Should be good to go now I think

[daxpedda]

Completely missed the changelog request! Sorry about that.

All good! I know how it is :)

Will wait for #108 before merging.

[stegaBOB]

Hey @daxpedda was wondering if there are any plans to cut a new release soon. We're currently working on a private registry so we aren't blocked on this or anything, but just wanted to ask since its been a while since the last one. Thanks! :)

Edit: just noticed the new PR... 🤦 that pretty much answers my question lol.