Enable mapping of field names with consumes argument for lightweight components

[RobbeSneyders]

The following code:

dataset = dataset.apply(
    "chunk_text",
    arguments={"chunk_size": 10, "chunk_overlap": 2},
    consumes={"text": "text_data"},
)


@lightweight_component(
    base_image="python:3.8",
    extra_requires=[
        # TODO: remove once we have a default base image
        "fondant[component]@git+https://github.com/ml6team/fondant@3f3d8ad",
    ],
)
class CalculateChunkLength(PandasTransformComponent):
    def transform(self, dataframe: pd.DataFrame) -> pd.DataFrame:
        dataframe["chunk_length"] = dataframe["chunk"].apply(len)
        return dataframe


_ = dataset.apply(
    ref=CalculateChunkLength,
    consumes={"chunk": "text"},
    produces={"chunk_length": pa.int32()},
)

leads to the following error:

fondant.core.exceptions.InvalidPipelineDefinition: Received a string value for key chunk in the `consumes` argument passed to the operation, but chunk is not defined in the `consumes` section of the component spec.

Raised at:

fondant/src/fondant/core/component_spec.py

Line 488 in abcd36f

raise InvalidPipelineDefinition(msg)