MlflowArtifactDataset.load() fails if artifact_path is not None and run_id is specified

[Galileo-Galilei]

Description

When you try to specify an artifact path and a run_id in an MlflowArtifactDataSet, you get an error.

This works:

mlflow_csv_dataset = MlflowArtifactDataSet(
    data_set=dict(type=CSVDataSet, filepath="path/to/df.csv"),
    artifact_path=None,
    run_id="1234",
)
mlflow_csv_dataset .load()

while this :

mlflow_csv_dataset = MlflowArtifactDataSet(
    data_set=dict(type=CSVDataSet, filepath="path/to/df.csv"),
    artifact_path="folder", # this is the difference
    run_id="1234",
)
mlflow_csv_dataset .load()

raises the following error: unsupported operand type(s) for /: 'str' and 'str':

[Galileo-Galilei]

The issue is still here when there is nested artifact_path: if the file does not exists, it is downloaded to self._filepath /artifact_path/filename.pkl and cannot be loaded (due to the artifact_path suffix)