fixed eval_model nb, added ZanjHookedTransformer.config ref

the `eval_model.ipynb` notebook has a function `testdata_plot_predicted_path`
which was using `model.zanj_model_config` to get the tokenizer, an attribute
missing from the `RandomBaseline` class since it only inherits from `HookedTransformer`

to fix this:

- `ZanjHookedTransformer` now has a `config` property which just
  accesses the `zanj_model_config` used by the parent `ConfiguredModel`
- `testdata_plot_predicted_path` now uses `model.config` everywhere

maze_transformer/training/config.py

@@ -568,6 +568,10 @@ def __init__(self, cfg_holder: ConfigHolder) -> None:
             tokenizer=cfg_holder.tokenizer,
         )
 
+    @property
+    def config(self) -> ConfigHolder:
+        return self.zanj_model_config
+
     def _load_state_dict_wrapper(
         self,
         state_dict: dict[str, Any],