fix: Fix loses associated bounds for replace_derive_with_manual_impl

[A4-Tacks]

Fixes #19878

Example

#[derive(Clo$0ne)]
struct Foo<T: core::ops::Deref>(T::Target);

Before this PR

struct Foo<T: core::ops::Deref>(T::Target);

impl<T: core::ops::Deref + Clone> Clone for Foo<T> {
    $0fn clone(&self) -> Self {
        Self(self.0.clone())
    }
}

After this PR

struct Foo<T: core::ops::Deref>(T::Target);

impl<T: core::ops::Deref + Clone> Clone for Foo<T>
where T::Target: Clone
{
    $0fn clone(&self) -> Self {
        Self(self.0.clone())
    }
}

[ShoyuVanilla]

This behaves correctly, but the variable name and passing it as the param with the same name might be misleading. We’re collecting clauses from the ADT here, not from the trait.

[ShoyuVanilla]

rust-analyzer/crates/syntax/src/ast/make.rs

Lines 324 to 337 in 9efc1f7

	pub fn impl_trait(
	attrs: impl IntoIterator<Item = ast::Attr>,
	is_unsafe: bool,
	trait_gen_params: Option<ast::GenericParamList>,
	trait_gen_args: Option<ast::GenericArgList>,
	type_gen_params: Option<ast::GenericParamList>,
	type_gen_args: Option<ast::GenericArgList>,
	is_negative: bool,
	path_type: ast::Type,
	ty: ast::Type,
	trait_where_clause: Option<ast::WhereClause>,
	ty_where_clause: Option<ast::WhereClause>,
	body: Option<ast::AssocItemList>,
	) -> ast::Impl {

We might need to improve the above API as well to accept iterator or sth for where clauses but I think this PR is good as is for now