Fix normal function call signature

The normal function should not take a reference to an i32, since the inverse function does not take a reference to T.

Giving these functions identical call signatures improves clarity, and it makes more sense to pass integers normally than by reference.

first-edition/src/traits.md

@@ -414,7 +414,7 @@ impl ConvertTo<i64> for i32 {
 }
 
 // Can be called with T == i32.
-fn normal<T: ConvertTo<i64>>(x: &T) -> i64 {
+fn normal<T: ConvertTo<i64>>(x: T) -> i64 {
     x.convert()
 }