feat(database): shape the library like files, and shrink the catalog

Plans 013 and 014, the album page that prompted them, and the smaller
fixes they turned up. Changelog, largest first.

## The local library is shaped like files, not like MusicBrainz

`audio_files` carries its own tags and points at `albums` and
`artists`; `file_genres` is the one real many-to-many. `recordings`,
`release_group_recordings`, `artist_credit`, `artist_credit_artist`,
`recording_genres`, `release_groups` and `release_to_rg` are gone from
the local side, and with them a six-way join in every read, a
`MIN(release_group_id)` subquery in eleven queries and a
first-credited-artist subquery in nine. Measured on a real 25,966-file
library, every many-to-many that model expressed was 1:1 in the data.

- Ownership is a file. `GetFilePathsByRecordingMBIDs`,
  `LibraryMBIDIndex.CheckMBIDs`, `collectLibraryEntities` and
  `pruneStaleLocalCrossReferences` all join `audio_files`, so the 812
  orphaned recordings, 216 release groups and 260 artists that library
  carried are now structurally impossible.
- One projection: every track query selects from the `track_metadata`
  view, one row type, one mapper. Nine hand-rolled copies had drifted
  far enough to report different years on different screens.
- `library_id = 0` means every library, so each list query exists once
  instead of scoped and unscoped with a branch at every call site.
- No migration chain. `sql/schemas/` is the one description of the
  shape; `sql/migrations/`, `applyMigrations` and `schema_migrations`
  are squashed away, along with the drift between them that had sqlc
  generating against a stale schema.
- `database.InsertTestTrack` is the one test seeder; twenty test files
  had been assembling the old FK chain each in its own order.

## The catalog stores its ids as bytes

`explore_index`'s three 36-char MBID columns and its entity-type text
are 16 raw bytes and a small integer. The table and its six indexes go
780 MB to 405 MB on a real 2,052,200-row catalog, which is why a fresh
install is ~0.6 GB rather than ~1.0 GB.

- `backend/explore/mbid.go` is the only place the encoding is known;
  everything above it speaks dashed strings.
- `CHECK(length(mbid) = 16)` makes a stringly write fail at the insert
  rather than silently returning no rows, since SQLite does not coerce
  between TEXT and BLOB.
- The importer asks the artifact what encoding it carries and converts
  on the way in, so the artifact already published keeps working and no
  format bump is needed.
- `indexRowColumns`/`scanIndexRow` replace four copies of a 22-column
  list, and `TestStoredEncodingRoundTrips` sweeps every read path.

## An album page that says how much of the album is yours

- One question, asked once: is there a file. `filePaths` is filled by a
  single batched lookup when the tracklist settles, and the badge, the
  Play count, the dimmed rows and every menu item read it — replacing
  four claims of decreasing confidence that could show a green tick on
  an album whose every action did nothing.
- Play, Play 7 of 12, or no play button at all.
- `total_tracks` on `explore_index` (~2 bytes over 400,677 release
  groups) and on `audio_files` from tags that have always carried it:
  a complete MBID-matched album now makes no catalog call at all, where
  it used to spend the most expensive request the app makes.
- A merged cluster shows the running order the most releases agree on,
  and the version list marks the release you own rather than standing a
  synthetic entry in for it.
- `AlbumReleasesFailed`: a slow fetch is no longer reported as a failed
  one by a 12-second timer.
- Rows not in the library are dimmed in place (with `aria-disabled`)
  instead of the owned ones wearing a green tick and a legend.

## Caches and cover art get ceilings

- Only the three tiers of a cover are stored; the full-resolution copy
  nothing rendered was 1,134 MB of a 1.4 GB covers directory.
- One artist portrait is downloaded and the rest are remembered as
  URLs — 4.1 GB of a 5.3 GB cache was candidates no code path reads.
- `browsedArtBudget` and `httpCacheBudget` bound what an age cannot:
  the same install held art for 5,770 artists in a 1,301-artist
  library.
- `OrphanedArtistImagesJob` joined a bare MBID onto a sharded
  directory, so it deleted the rows that were the only record of the
  files it left behind. `explore.ArtistImageDir` is that layout's one
  definition now.

## The autotag queue asks whether there is work

`tagging_items` was a row per album folder, not a queue, and no query
read the `tag_status` column that held the answer. The four queue
queries ask the files, which matters most where it is least visible:
`startPrefetch` was scoring every album in a tagged library against
MusicBrainz.

## Phantom playlist tracks resolve in place

An M3U8 imported before its files leaves phantom rows; they now match
by path and fall back to position, keep their place in the playlist
when resolved, and pair best-first so two phantoms cannot claim the
same file.

## Playing a track plays the list it is in

Double-click, and Play on a single row's menu, queue the list as
displayed with `startIndex` on that row — the album page and the track
list used to queue one track and discard the album around it. A
multi-row selection still plays exactly itself.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AfVYUVExXsx1nSWrXN8mAh

backend/database/sql/schemas/audio_files.sql

@@ -1,34 +1,92 @@
+-- One row per audio file, and the file's tags live on it.
+--
+-- This table used to be a stub — path, format, a foreign key — with
+-- every tag-derived field one join away in `recordings`, which was in
+-- turn linked to an album through `release_group_recordings` and to an
+-- artist through `artist_credit` + `artist_credit_artist`.  That is
+-- MusicBrainz's data model, and it is the right model for MusicBrainz:
+-- a recording really can appear on many releases and a credit really
+-- can list many artists.
+--
+-- It was the wrong model here, and the library said so.  Measured on a
+-- real 25,966-file library: **no** recording had more than one file,
+-- **no** recording belonged to more than one release group, and 3 of
+-- 2,823 credits listed more than one artist.  Every many-to-many the
+-- schema modelled was 1:1 in the data, and the cost of modelling it
+-- anyway was a six-way join in every read, a `MIN(release_group_id)`
+-- subquery in eleven queries to collapse a fan-out that never happened,
+-- a first-credited-artist subquery in nine more to collapse the other
+-- one, and — the reason this changed — a whole class of bugs where a
+-- `recordings` row **outlived the file that created it**.  Retagging a
+-- file created a new recording and abandoned the old one, so the same
+-- library carried 812 recordings, 216 release groups and 260 artists
+-- with no file behind them, and everything that asked "do I own this"
+-- by looking for a metadata row got 129 confident yeses for tracks
+-- that could not be played.
+--
+-- With the tags on the file, ownership is not a rule anyone can forget:
+-- the row *is* the file.
 CREATE TABLE IF NOT EXISTS audio_files (
-  id integer PRIMARY KEY,
-  file_path text NOT NULL UNIQUE,
-  length_milliseconds int NOT NULL,
-  file_type_id int NOT NULL,
-  recording_id int NOT NULL,
-  sample_rate int NOT NULL DEFAULT 0,
-  bit_depth int NOT NULL DEFAULT 0,
-  channels int NOT NULL DEFAULT 0,
-  bitrate int NOT NULL DEFAULT 0,
-  file_size int NOT NULL DEFAULT 0,
-  basename text NOT NULL DEFAULT '',
-  library_id int NOT NULL DEFAULT 0,
-  play_count int NOT NULL DEFAULT 0,
-  last_played datetime,
-  tag_status TEXT NOT NULL DEFAULT 'untagged'
+  id                 INTEGER PRIMARY KEY,
+  file_path          TEXT NOT NULL UNIQUE,
+  library_id         INTEGER NOT NULL DEFAULT 0,
+  file_type_id       INTEGER NOT NULL,
+
+  -- Audio properties, read from the file itself.
+  length_milliseconds INTEGER NOT NULL,
+  sample_rate        INTEGER NOT NULL DEFAULT 0,
+  bit_depth          INTEGER NOT NULL DEFAULT 0,
+  channels           INTEGER NOT NULL DEFAULT 0,
+  bitrate            INTEGER NOT NULL DEFAULT 0,
+  file_size          INTEGER NOT NULL DEFAULT 0,
+
+  -- Tags.  `artist_credit` is the credit as tagged ("A feat. B") and is
+  -- for display; `artist_id` is the primary artist it resolves to, and
+  -- is what grouping, browsing and the artist page use.  Keeping both
+  -- is what makes the credit table unnecessary: the string is the only
+  -- thing that was ever read off it.
+  title              TEXT NOT NULL DEFAULT '',
+  artist_credit      TEXT NOT NULL DEFAULT '',
+  artist_id          INTEGER,
+  album_id           INTEGER,
+  track_number       INTEGER,
+  disc_number        INTEGER,
+  -- The denominator the tag declared: the 12 in "5/12", per disc.  It
+  -- is what lets "do I have all of this album" be answered from disk
+  -- instead of from MusicBrainz.  NULL means the tag did not say, which
+  -- is a third state and not the same as zero.
+  total_tracks       INTEGER,
+  year               INTEGER,
+  composer           TEXT NOT NULL DEFAULT '',
+  comment            TEXT NOT NULL DEFAULT '',
+  recording_mbid     TEXT,
+
+  -- Library bookkeeping.
+  basename           TEXT NOT NULL DEFAULT '',
+  group_key          TEXT NOT NULL DEFAULT '',
+  -- File mtime as a Unix timestamp in seconds, captured at import and
+  -- compared against the on-disk mtime during a scan to detect files
+  -- another application retagged in place.
+  modified_at        INTEGER NOT NULL DEFAULT 0,
+  play_count         INTEGER NOT NULL DEFAULT 0,
+  last_played        DATETIME,
+  tag_status         TEXT NOT NULL DEFAULT 'untagged'
     CHECK(tag_status IN (
       'untagged', 'auto_matched', 'user_confirmed', 'user_skipped_permanent'
     )),
-  group_key TEXT NOT NULL DEFAULT '',
-  -- File mtime as a Unix timestamp in seconds, captured at import.
-  -- Compared against the on-disk mtime during a scan to detect files
-  -- another application retagged in place.  0 means "never recorded"
-  -- (rows predating migration 47) and is treated as not-stale so an
-  -- upgrade does not re-import the whole library.
-  modified_at int NOT NULL DEFAULT 0,
+
   FOREIGN KEY(file_type_id) REFERENCES file_types(id),
-  FOREIGN KEY(recording_id) REFERENCES recordings(id),
-  FOREIGN KEY(library_id) REFERENCES libraries(id)
+  FOREIGN KEY(library_id)   REFERENCES libraries(id),
+  FOREIGN KEY(artist_id)    REFERENCES artists(id),
+  FOREIGN KEY(album_id)     REFERENCES albums(id)
 );
 
+CREATE INDEX IF NOT EXISTS idx_audio_files_album_id
+    ON audio_files(album_id);
+
+CREATE INDEX IF NOT EXISTS idx_audio_files_artist_id
+    ON audio_files(artist_id);
+
 CREATE INDEX IF NOT EXISTS idx_audio_files_basename
     ON audio_files(basename);
 
@@ -36,10 +94,17 @@ CREATE INDEX IF NOT EXISTS idx_audio_files_group_key
     ON audio_files(group_key) WHERE group_key != '';
 
 CREATE INDEX IF NOT EXISTS idx_audio_files_library_id
-  ON audio_files(library_id);
+    ON audio_files(library_id);
 
-CREATE INDEX IF NOT EXISTS idx_audio_files_recording_id
-    ON audio_files(recording_id);
+-- The ownership question, asked by MBID: "is there a *file* with this
+-- recording MBID".  Nothing may answer it from a metadata table again.
+CREATE INDEX IF NOT EXISTS idx_audio_files_recording_mbid
+    ON audio_files(recording_mbid) WHERE recording_mbid IS NOT NULL;
+
+-- Answers "does this tagging group still contain untagged files" in one
+-- seek per group.  The autotag queue asks it once per row.
+CREATE INDEX IF NOT EXISTS idx_audio_files_untagged_group_key
+    ON audio_files(group_key) WHERE tag_status = 'untagged';
 
 CREATE INDEX IF NOT EXISTS idx_audio_files_tag_status_untagged
     ON audio_files(library_id) WHERE tag_status = 'untagged';