dask_ml does not work with the "k-means++" method on version 2024.4.4.

With the latests versions of dask_ml, `KMeans` and  `k_init` try to use
the `kmeans_plusplus` of sklearn without the `sample_weights` argument.

Rechunk the data array to keep the same results of k_init.

meta(deps): add bob as dependency in new structure

See merge request !77

meta [readme]: Switch  the readme.rst to markdown

See merge request !76

[pyproject.toml] Changing documentation link to master/sphinx

See merge request !75

Set the factor analysis score functions to always return a float

See merge request !74

Update ISV to accept probe templates. Update QA.

See merge request !73

Remove bob.extension from dependencies

See merge request !72

[fix] Specify the chunk size of arrays for k_init.

Closes #49

See merge request !71

The "auto" chunk size somehow breaks k_init for large numpy arrays.

Cookiecutter restructuration

Closes #47

See merge request !69