RUF012 triggers many false positives (are they really? they are correct) in some projects

Hello 👋🏽

In this [PR](https://github.com/astral-sh/ruff/pull/4390) RUF012 was extended to apply to non-dataclass classes.

This has the (maybe desirable, maybe not, but was surprising to me) effect of _forcing typing_ for a RUF rule, in places where you would not necessarily use typing (if you did not want to).

For instance, in Django Rest Framework ([from their docs](https://www.django-rest-framework.org/api-guide/serializers/#modelserializer)):

```python
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
fields = ['id', 'account_name', 'users', 'created'] # RUF012
```

Or in Django [itself](https://docs.djangoproject.com/en/4.2/ref/models/options/#django.db.models.Options.indexes):

```python
class Customer(models.Model):
first_name = models.CharField(max_length=100)
last_name = models.CharField(max_length=100)

class Meta:
indexes = [
models.Index(fields=["last_name", "first_name"]), # RUF012
models.Index(fields=["first_name"], name="first_name_idx"), # RUF012
]
```

You could argue that's good, but at least to me, it's _different_ from dataclasses, since in there you are required to use typing to even use them, so they are opt-in even before you use ruff. You could also argue both Django and Django Rest Framework should be recommending tuples instead.

Isn't this too trigger happy for a RUF rule? Does it make sense to split it from the dataclass one?

EDIT: Actually, it is a separated code already 🤦🏽

In any case, feel free to close if this is just what it is! Ignoring the rule is easy for these kind of projects.