Enable caching for write components

[PhilippeMoussalli]

Write components do not write any new manifests since they don't alter the data. However, the cache estimation is based on checking for existing manifests based on the cache key. We need to somehow enable this either by writing a dummy manifest (propagating the previous one) or handling the logic separately.

Reproducible example:

import dask.dataframe as dd
import pandas as pd
import pyarrow as pa

from fondant.component import (
    DaskLoadComponent,
    DaskWriteComponent,
    PandasTransformComponent,
)
from fondant.pipeline import Pipeline, lightweight_component

pipeline = Pipeline(
    name="all-bio-is-just-chemistry",
    base_path="/Users/georgeslorre/ML6/internal/express/sweet-pipelines/pure/artifacts",
)


@lightweight_component(
    base_image="fndnt/fondant-base:dc8d9b208802ee85796238ff9a903c8bf66906e8-python3.8"
)
class CreateData(DaskLoadComponent):
    def load(self) -> dd.DataFrame:
        df = pd.DataFrame(
            {
                "x": [1, 2, 3],
                "y": [4, 5, 6],
            },
            index=pd.Index(["a", "b", "c"], name="id"),
        )
        return dd.from_pandas(df, npartitions=1)


dataset = pipeline.read(
    ref=CreateData,
    produces={"x": pa.int32(), "y": pa.int32()},
)


@lightweight_component(
    base_image="fndnt/fondant-base:dc8d9b208802ee85796238ff9a903c8bf66906e8-python3.8"
)
class FeatOne(PandasTransformComponent):
    def transform(self, dataframe: pd.DataFrame) -> pd.DataFrame:
        dataframe["one"] = dataframe["x"].map(lambda x: x + 1)
        return dataframe


f1 = dataset.apply(
    ref=FeatOne,
    produces={"x": pa.int32(), "y": pa.int32(), "one": pa.int32()},
    consumes={"x": pa.int32(), "y": pa.int32()},
)


@lightweight_component(
    base_image="fndnt/fondant-base:dc8d9b208802ee85796238ff9a903c8bf66906e8-python3.8"
)
class WriteOut(DaskWriteComponent):
    def write(self, dataframe: dd.DataFrame) -> None:
        dataframe.to_parquet("./artifacts/out")


_ = f1.write(
    ref=WriteOut,
    consumes={"x": pa.int32(), "y": pa.int32(), "one": pa.int32()},
)