feat: write public stats to our PG DB

[bajtos]

A prototype implementing code for producing stats to power our Spark Public Dashboard. Design doc:
https://pl-strflt.notion.site/DB-for-SPARK-public-data-a79cbb7a0ff5447980b714afdb2645d6

1. Add a runtime dependency on a PG database, using a new database
   name spark_stats (originally: spark_public). Setup database schema migrations executed
   at service startup.

2. Define the first table retrievals to keep stats about all vs
   succesfull retrievals performed each day.

3. Update the evaluate step to update these stats for each round
   completed.

TODO

• Setup PG DB for GitHub Actions to make the tests pass
• Create a new PG user spark_evaluate and a new database spark_stats
• Define DATABASE_URL in Fly.io secrets

Links:

• Public Spark Dashboard MVP space-meridian/roadmap#57

[bajtos]

Last minute change: let's rename the DB from spark_public to spark_stats. It better captures what we are storing. In the future, we can store in this DB data for our private dashboards too.

[juliangruber]

Great work!