Denormalize library.artwork_url onto flowsheet on bin-pick

[jakebromberg]

Summary

• The bin-pick branch of addEntry already calls getAlbumFromDB to backfill album info, but did not pull library.artwork_url even though the column has existed since migration 0045. This patch adds it to the SELECT so the freshly-inserted flowsheet row carries artwork at INSERT time.
• Eliminates the metadata race for bin-picks on every client surface at once: the addEntry HTTP response carries the URL to dj-site (which polls every 60s and won't otherwise see it for up to that long), and the playlist-proxy's synchronous SELECT artwork_url WHERE … on the tubafrenzy created event reads a non-null value.
• Free-form entries (no album_id) still rely on the asynchronous LML enrichment UPDATE — the second-emit + in-process proxy re-enrich tracked in the rest of Re-emit liveFs SSE event after metadata enrichment UPDATE lands #628 covers that path.

Part of #628 (does not close it; the free-form-entry second-emit fix is the remainder).

Test plan

• tests/unit/controllers/flowsheet.controller.test.ts: new tests for the populated and null library.artwork_url cases on the bin-pick path. All 36 tests in that file pass.
• prettier --check on the touched files.
• After merge, run SELECT count(*) FILTER (WHERE artwork_url IS NOT NULL), count(*) FROM library against prod to confirm population is high enough that this patch does meaningful work in practice. (If population is low, this still doesn't regress — null artwork_url passes through and the LML enrichment path takes over as before.)

[jakebromberg]

Holding this in draft pending the library-side artwork backfill.

Per the precondition comment on #628 (issuecomment-4338014422), the production query reported there shows that library.artwork_url is populated for only 155/64163 rows (0.2%) today.

The denormalization in this PR is correct and harmless, but until the library-side backfill ships, it denormalizes ≈zero rows in practice — the user-visible win is near zero. Filing the backfill as its own follow-up; will rebase + un-draft this once that has run and the population number is materially higher.

In the meantime, surface 2 in #628 (second-emit + in-process proxy re-enrich) is the load-bearing next step and will be picked up in a separate session.