Fix ptr_arg suggests changes when it's actually better not to bother

[ghost]

No longer suggests &[i32] or &mut [i32] instead of &Vec<i32> or &mut Vec<i32> (also: Path and PathBuf, etc.) for the parameter type when the parameter name starts with an underscore (or, if that does not start with one, then a local let binding in the function body, pointing to the same value, does) – (Fixes #13489, fixes #13728)

changelog: fix false positive: [ptr_arg] no longer triggers with underscore binding to &mut argument
changelog: fix false positive: [ptr_arg] no longer triggers with underscore binding to &T or &mut T argument
Edit: This change has been extended to all references, not just mutable ones. See discussion below.

[rustbot]

r? @Jarcho

rustbot has assigned @Jarcho.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[ghost]

This change only makes an exception for mutable all reference types having binding names with an underscore prefix.

An exception is not made in the following scenarios:

• _x: &Vec<i32> – It's not considered okay because requiring a reference to Vec to be passed should never be necessary. In this case, let's still suggest _x: &[i32] despite that the parameter is intentionally unused.
  Note: Maybe, when the underscore is present, modify the hint to suggest either &[i32] or &mut Vec<i32>?
  Edit: The underscore named parameter is a good indicator that the signature matters for some reason.
• x: &mut Vec<i32> (or x: &Vec<i32>), but when x is not used – It may or may not be considered okay. However, such a case is outside the scope of this change.
  Note: Maybe decide whether, in general, an unused warning should inhibit ptr_arg lints, and open a new issue?

[Jarcho]

I would ignore all references, not just mutable ones. An underscore named parameter is a good indicator that the signature matters for some reason (e.g. for function pointers).

[ghost]

For example, it matters for methods that are not otherwise possible to call, or the user may want to clone the value.

I've applied the necessary changes. I've also updated the parameter names in some earlier tests so that they remain effective.

[Jarcho]

Looks good. Can you just squash the two commits please.

[ghost]

Sure, there you go, @Jarcho