Streaming implementations of post processing

[WardLT]

We are going to start dealing with large scale datasets, and we'll need to rewrite our post processing code to work on batches or streams of time series data.