Fixed type (pangeo-forge#483)

docs/pangeo_forge_recipes/tutorials/xarray_zarr/multi_variable_recipe.ipynb

@@ -2017,7 +2017,7 @@
     "We will write a Beam transform that fixes both these issues. This is achieved by creating a [composite transform](https://beam.apache.org/documentation/programming-guide/#composite-transform-creation), which is a subclass of the `apache_beam.PTransform` class that overrides the `expand()` method to specify the actual processing logic. \n",
     "\n",
     "In this example:\n",
-    "* `expand()` operates on an `apache_beam_PCollection`, performing a one-to-one mapping using [`apache_beam.Map`](https://beam.apache.org/documentation/programming-guide/#pardo) of input elements to output elements, i.e. for each input element, it applies `fix_encoding_and_attrs()` that produces exactly one output element.\n",
+    "* `expand()` operates on an `apache_beam_PCollection`, performing a one-to-one mapping using [`apache_beam.Map`](https://beam.apache.org/documentation/programming-guide/#pardo) of input elements to output elements, i.e. for each input element, it applies `_fix_encoding_and_attrs()` that produces exactly one output element.\n",
     "* As the preprocessor transform will be preceded by the `OpenWithXarray` transform in the pipeline, each input collection element will be a `pangeo_forge_recipes.transforms.Indexed[T]`. In this case each tuple will contain an index and an `xarray.Dataset`. The output tuple will contain the original index and the preprocessed `Dataset`."
    ]
   },