Rule idea: flag literal attributes in monkeypatching/mocking APIs

### Summary

This is a rough idea that probably needs refining/discussion 🙂

## Overview

Pytest and others expose monkeypatch/mocking APIs like:

```python
monkeypatch.setattr(some_module, "my_important_api", replacement)
```

(and same for `delattr`.)

When called, this temporarily monkeypatches `some_module.my_important_api` with `replacement` (which could be an object, callable, etc.).

## Problem statement

The above works, but has a footgun: if the user performs a refactor (e.g. from `my_important_api` to `my_critical_api`), then may miss this monkeypatch (since it's a string literal, not a usage/reference). This is particularly likely to happen with IDE/LSP refactors, since syntax-aware rename-all functionality can't look at these kinds of non-type string literals by design.

The impact of this depends: sometimes users notice it immediately (since their monkeypatch stops working), while other times it makes tests subtly incorrect or unreliable in ways that aren't noticed until much later.

## Solution statement

In general, users should _probably_ write:

```python
monkeypatch.setattr(some_module, some_module.my_important_api.__name__, replacement)
```

That'll behave the same as above, _but_ will ensure that refactors/mechanical renames propagate as expected.

## Problems

1. This won't work with dynamic attributes; not sure if this is a problem or not.
2. Using `__name__` is arguably pretty unsightly and is maybe considered unidiomatic, not sure 🙂

## Alternatives

1. Do nothing.
2. Maybe Pytest and others with monkeypatching APIs should support functions as first-class "nameable" objects? In other words, Ruff could instead recommend:

```python
monkeypatch.setattr(some_module, some_module.my_important_api.__name__, replacement)
```

...and Pytest would do the `__name__` access internally. This would require upstream changes, but would be a lot visually cleaner from the user's side.

## Other references

xref https://github.com/astral-sh/ruff/issues/12850 for another (but I think unrelated) mocking/monkeypatching feature request.

Start by reviewing the proposed monkeypatch.setattr and delattr examples and the related issue xref, then clarify which APIs and dynamic-attribute cases are in scope. Done means the rough idea has a settled rule behavior and implementation scope that can be picked up without further design discussion.

Written by the indexing model from the issue text.