Consolidate the mapping data into the MTTL

There are two major problems with our MTTL data flow from external repos to the MTTL from the pipeline to populate metrics. The problems are explained in the google slides presentation here.

This issue will also require the introduction of a mongodb service that will be used in the pipeline. The database tools that this will provide will help us streamline the scripts we are currently using in the pipeline.

AC

• redesign the 'training' and 'eval' fields in the KSAT json to be easier to use
• migrate all external rel-link info into newly designed 'training' and 'eval' fields
• rework all pipeline scripts that require interaction with the datasets to use new mongodb service
• create documentation for new rel-link, training, eval question workflow