Timeseries reconciliation

[intarga]

Make a script to help content managers find and combine duplicate timeseries from different sources

[ketilt]

I could probably do this. Would this make more sense as an integral part of LARD, or located somewhere else in the architecture?

[intarga]

I think it makes sense to do this in the ingestor (though isolated enough that we can break it out later), since that already runs an HTTP server and has a database connection pool.

I suggest breaking down the task like so:

• Informally set out the algorithm for reconciliation on paper, defining a priority order for the data sources
• Write a SQL query that finds reconciliation candidates, and another that resolves them
• Design a web UI
• Write some plumbing handler functions in the ingestor to connect the web UI to the queries
• Write some test cases to ensure this works as expected

Please reach out for help at any point, I imagine especially that tasks 4 and 5 will go a lot smoother with some help from Manuel or I. We can also probably offer useful insight on the structure of the tables for task 2