fix(datasets): Add parameter to enable/disable lazy saving for PartitionedDataset

[ElenaKhaustova]

Description

Fixes #759

Cause of the issue: #759 (comment)

Dataset lazy saving docs: https://docs.kedro.org/en/stable/data/partitioned_and_incremental_datasets.html#partitioned-dataset-lazy-saving

Development notes

1. Added a parameter (save_lazily) to enable/disable lazy saving for PartitionedDataset. The parameter is enabled by default but users still need to wrap objects with callable as it was required before to apply lazy saving, so the change is not breaking.
2. We didn't add a similar argument for save function as was suggested initially since it contradicts the definition of AbstractDataset.save
3. Added callable object saving and loading test
4. Following docs update PR: Update partitioned dataset lazy saving docs kedro#4402

How to test without TensorFlow

The following code will execute without errors and save two objects if save_lazily is set to False. The code will fail and TestSaveCallable.__calll__ will be called if save_lazily is set to True or not set.

from kedro_datasets.partitions import PartitionedDataset


class TestSaveCallable:
    def __init__(self, a: int):
        print(f"--- Initialized {a} ---")

    def __call__(self, *args, **kwargs):
        print("--- Called ---")


class TestSaveNotCallable:
    def __init__(self, a: int):
        print(f"--- Initialized {a} ---")


def main():
    partitioned_dataset = PartitionedDataset(
        path="data/01_raw/pickle",
        dataset="kedro_datasets.pickle.PickleDataset",
        filename_suffix=".pkl",
        save_lazily=False
    )

    save_dict = {
        "object_1": TestSaveNotCallable(1),
        "object_2": TestSaveCallable(2),
    }

    partitioned_dataset.save(save_dict)


if __name__ == "__main__":
    main()

Checklist

• Opened this PR as a 'Draft Pull Request' if it is work-in-progress
• Updated the documentation to reflect the code changes
• Updated jsonschema/kedro-catalog-X.XX.json if necessary
• Added a description of this change in the relevant RELEASE.md file
• Added tests to cover my changes
• Received approvals from at least half of the TSC (required for adding a new, non-experimental dataset)

[noklam]

I think callable data (that is not intend to be lazy saved) is rare and is not worth to break PartitionedDataset for it. Left the comment in the original issue since I saw two PRs are associated.

#759 (comment)

[DimedS]

Thanks for the PR, @ElenaKhaustova ! LGTM!

[merelcht]

Nice and clean solution 👌

[Galileo-Galilei]

I really like the solution, but a bit confused about the naming choice which I find inconsistent with other datasets. What's the rationale of not using save_args and add save_lazily a key of the dictionary?