`Pipeline` is not designed to be extended

[roygutg]

Currently, behavioral/Pipeline.py contains only one Pipeline class, intended to download data from RedMetrics1, using behavioral.Downloader.
We would want to enable using pipelines with different data sources, e.g., from file, downloaded from AWS S3, or downloaded from RedMetrics2.

To that end, the current Pipeline class should split into two: (1) an abstract base class, naive to its raw_data source, and (2) a RedMetircs1Pipeline (or similar), extending the abstract Pipeline by providing raw data from RM1 using the appropriate Downloader.