[refactor] Simplify handling of trait bounds and aliases

[MarijnS95]

All the cleanups from #1153 separated in nicely documented, individually reviewable commits.

This includes a simplification of where T: IsA<Foo> into specifying the trait bound directly when the type parameter is defined: func<T: IsA<Foo>>.

As requested by @GuillaumeGomez the "fix" for clippy::use_self is in here too, but in hindsight I'm not too sure how useful it is. This is what it does:

diff --git a/gdk/src/auto/device.rs b/gdk/src/auto/device.rs
index f53e2a154f..9994f88311 100644
--- a/gdk/src/auto/device.rs
+++ b/gdk/src/auto/device.rs
@@ -450,7 +450,7 @@ impl Device {
     }

     #[doc(alias = "changed")]
-    pub fn connect_changed<F: Fn(&Device) + 'static>(&self, f: F) -> SignalHandlerId {
+    pub fn connect_changed<F: Fn(&Self) + 'static>(&self, f: F) -> SignalHandlerId {
         unsafe extern "C" fn changed_trampoline<F: Fn(&Device) + 'static>(
             this: *mut ffi::GdkDevice,
             f: glib::ffi::gpointer,

F of the function and F of the unsafe extern "C" fn still mean the same thing, though they are now textually different. The the nested unsafe extern "C" fn cannot reference type parameters (Self is a type parameter!) of the surrounding context, which is why it still carries Device instead of Self.
A bit further don this code-block the pointer of changed_trampoline::<F> is taken, to pass type parameter F of the surrounding function into the nested one (hence they reference the same type).

[GuillaumeGomez]

So I guess if we regen with this, the only changes we should see are the Self ones. If so, looks good to me! :)

[GuillaumeGomez]

Thanks!