Demonstrate the "udefinert stor datasamling" requirement with numbers. - [ ] run `ALBUM_COUNT=100000 pnpm db:seed:large` (and ideally 1M) on the VM - [ ] record `EXPLAIN ANALYZE` for a search, a filtered search and a deep page - [ ] show the GIN index is used (not a sequential scan) - [ ] write the results up in README under a "Scalability" heading **Acceptance:** the README shows before/after index timings and the page of query plans.