Geoprocessing projects use precalculated metrics mainly for the denominator in calculating % overlap with sketches.
Since the datasets are so big, the methods for precalculating may vary, via custom scripting.
The end format should be metrics that are published, either in the repo, or in S3 so that they can be used internally or by external geoprocessing projects.
Consider adding a project client method to get precalculated metrics for external datasources