[dask] Dask estimators not supported on Windows

[Fish-Soup]

Description

As seen in 5918 dask training doesn't work on windows with a local cluster.

Reproducible example

    cluster = LocalCluster()  # Fully-featured local Dask cluster
    client = cluster.get_client()

    X, y = make_classification(n_samples=10000, n_features=20, n_informative=15, random_state=42)

    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

    chunk_size = 1000
    X_train_dask = da.from_array(X_train, chunks=(chunk_size, X_train.shape[1]))
    y_train_dask = da.from_array(y_train, chunks=(chunk_size,))

    dask_lgbm = lgb.dask.DaskLGBMClassifier(
        num_leaves=31,
        learning_rate=0.05,
        n_estimators=100,
        #machines=machines_str,
        client=client
    )

    dask_lgbm.fit(X_train_dask, y_train_dask)

    client.close()

Environment info

LightGBM 4.6.0

pip install lightgbm

Additional Comments

[jameslamb]

Thanks for using LightGBM.

Exactly as we said on #5918, which you linked:

We also already have a feature request in the project backlog to add Dask support on macOS and Windows: #3782

ref: #5918 (comment)

So we do not need this additional issue tracking that feature request. You can subscribe to #3782 to find out if/when support for the Dask estimators on Windows is added.

If you simply opened this issue to say "+1, please add this", just add a 👍🏻 reaction on #3782.

If you have specific information that'd change contributors' understanding of how valuable this feature is (for example "I work in a large research lab and we use Dask on Windows, not having this is preventing us from using LightGBM for our research"), comment on #3782 with that information.

If you'd like to see this support added and are interested in working on it, comment on #3782 and say that and we'd be happy to help you with the contribution process.