Would be nice to have more Pandas rolling window functions in streamz. I implemented rolling correlation and exponential weighted window. The EWM uses a cut-off threshold for finite window sizes. Is there interest in a PR of either?