Fix PydanticObjectId types being parsed as strings when using Pydan…

…tic's `model_validate_json`

Unsure if this could be fixed in a more stylish fashion, but it was simply incorrectly setup before, at least now it works. 

**Please make as many changes as needed if this solution could be improved.**

I believe this bugfix is quite important. If you parse an ID (let's call the variable `identifier`) from a JSON file in order to look it up on your Mongo database... you won't find it.

```python
 await cls.find_one(cls.id == identifier)
```
will return `None` even if the object exists, because `identifier` will be of type `str` instead of `ObjectId` when parsing from a JSON file with Pydantic's `model_validate_json`. 

Fixes BeanieODM#940.

beanie/odm/fields.py

@@ -159,7 +159,7 @@ def __get_pydantic_core_schema__(
         ) -> CoreSchema:  # type: ignore
             return core_schema.json_or_python_schema(
                 python_schema=plain_validator(cls.validate),
-                json_schema=str_schema(),
+                json_schema=plain_validator(cls.validate),
                 serialization=core_schema.plain_serializer_function_ser_schema(
                     lambda instance: str(instance), when_used="json"
                 ),