design stream schema and process for the intermediate and final representations of trialist user data

[jshslsky]

When a user's trial finishes, an analyzable data set needs to be generated. This analyzable data set slots the survey responses into buckets based on the randomly generated AB protocol, the number of cycles and the period length for each A or B period.

Marc S will write a DPU in Clojure to create the analyzable data set. Chris from Brown will write a DPU in R to create the final visualizable data set.

The workflow is as follows:

• An early morning job will run in ohmage to see if any participants have finished a trial on the previous day.
• For each participant that finished, their responses will be packaged up and passed to Marc's DPU.
• Marc's DPU will return the analyzable representation of the data and ohmage will store it as stream data.
• ohmage will then pass the analyzable data set to the R DPU (hosted in OpenCPU). This DPU could run for a potentially long time. Once this DPU returns, ohmage will store the final data analysis as a stream.
• The final results visualization will retrieve this final data representation from the ohmage stream API.

[jshslsky]

This has been done and the observer_stream definitions are deployed to opilots.