search: build and use a smaller cirrus index dataset
Build a parquet version of the cirrus index dataset without content fields. This task is added to the main dag. Switched the incoming_links job to use this new dataset. This means that we have to be a bit careful when deploying this patch to make sure that it's not deployed in the middle of a an import_cirrus_indexes run.
Depends on: repos/search-platform/discolytics!26 (merged)