Flag to skip centering in PCA

Hey @TomAugspurger, how opposed would you be to introducing small deviations from the scikit-learn interface to PCA like adding a flag in dask-ml.PCA to avoid requiring this centering?

https://github.com/dask/dask-ml/blob/b94c587abae3f5667eff131b0616ad8f91966e7f/dask_ml/decomposition/pca.py#L270-L271

That would certainly be a bad default, but if I know that is going to happen downstream from a scaling operation in a pipeline that already does the centering, I would like to have the option to avoid the extra operation. Is that reasonable or are you trying maintain something closer to perfect consistency?

Start with dask_ml/decomposition/pca.py at the linked lines 270-271 and review how PCA currently performs centering. Compare the requested option with the scikit-learn PCA interface and determine the API behavior and safeguards needed; the work is done when the proposed flag has an agreed design and its behavior is covered by the relevant PCA tests.

Written by the indexing model from the issue text.