Remove trait bounds on structs

[ethanfrey]

These are not needed in the structs, but rather in the impl blocks.
The stemmed from when I was newer to Rust and didn't realise that #[derive(Clone)] was valid if some generics didn't implement Clone. I discovered this is like impl<C: Clone> Clone for Msg<C> { ... }, so we get that check for free.

When working with multi-test and more powerful use of generics and traits, these restrictions were quite troublesome. If you have a type that doesn't implement one and it is used somewhere that needs it, that will cause a compiler error. So no need to enforce that on the type.

[uint]

For the record, the fact std library derives like #[derive(Clone)] require all type parameters to implement the trait is a bug.

rust-lang/rust#26925
https://www.reddit.com/r/rust/comments/517rsk/deriving_copy_for_a_struct_with_references_to/

Enforcing a trait bound on the types themselves is a valid thing to do and often useful, it just requires a manual impl block currently or to use a better derive macro from an external crate.

Edit: I'm not suggesting a different course of action, just adding info.

[hashedone]

Love the change.

A hint for implementing traits for generics - just use derivative - it saved me hundreds lines of code while implementing Default when not all fields has their default, and also Debug when not all fields are implementing it, but it is also helpful for reducing bounds (just by #[derivative(Clone(bound = ""))]). One of my fav utility crates.

[uint]

lgtm

[ethanfrey]

Yay! 🎉