refactor(download): rename Want/Request to Request/Download, unify downloads flow, add auto-download guardrails

The durable "I asked for this" record was called Want, and the one-shot
search-and-grab attempt was called Request — names that didn't match
what either actually did. Want is now Request, and the old Request/Item
is now Download/DownloadItem, with a table-rename migration
(download_wants -> download_requests, old download_requests ->
download_downloads) safe against both fresh installs and existing data.

Every anchored manual download now upserts/reuses a durable Request
before running, so a "download now" that finds nothing is picked up by
the background reconciler automatically instead of just failing with
no trace — the gap that caused this session's repeated "no candidates
found" failures on the same album.

Also adds auto-download guardrails (file-size min/max with a preferred
target, allowed file types) that gate what the pipeline may grab
unattended, live-editable from a new settings section. The frontend's
wanted-view becomes downloads-view, with a new Downloads tab showing
attempt/transfer history that previously had no UI at all.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y2Agd9af5hE7qzti2ackiS

backend/database/sql/schemas/download_requests.sql

@@ -1,49 +1,79 @@
--- One row per "go find me this", from the moment the user asks until
--- the files are in the library or the attempt is abandoned.
+-- A Request is a persistent "I want this", stored as a MusicBrainz ID
+-- and almost nothing else.  It outlives every download attempt made on
+-- its behalf: nothing being findable today is the normal case for
+-- obscure music, and the correct response is to try again next week,
+-- not to show the user a failed row they have to remember to retry.
 --
--- release_mbid / release_group_mbid are the anchor: a request that
--- carries one can be matched against a known tracklist at import time,
--- which is what makes unattended completion safe.  Free-text requests
--- (both NULL) are always presented to the user for confirmation.
---
--- `expected` caches the anchor's tracklist as JSON so ranking and
--- import do not have to re-resolve it, and so a request survives the
--- explore index being rebuilt underneath it.
+-- Because a Request is only an MBID, it stays true when everything
+-- around it changes: the explore index is rebuilt, a provider is
+-- swapped out, the release the user originally saw is superseded by a
+-- remaster.  The display fields are a cache for the list view and are
+-- never consulted for matching.
 
 
 CREATE TABLE IF NOT EXISTS download_requests (
-    id                 TEXT PRIMARY KEY,
-    library_id         INTEGER NOT NULL,
-    -- source records where the request came from: 'explore-album',
-    -- 'explore-artist', 'missing-album', 'wanted', 'manual'.
-    source             TEXT    NOT NULL DEFAULT 'manual',
-    -- want_id is set when the reconciler raised this request from the
-    -- wanted list, so the outcome can be written back to the want.
-    -- NULL for one-off requests the user started by hand.  Requests are
-    -- disposable and wants are not, so the delete is a SET NULL rather
-    -- than a cascade in either direction.
-    want_id            INTEGER REFERENCES download_wants(id) ON DELETE SET NULL,
-    release_mbid       TEXT,
-    release_group_mbid TEXT,
-    -- recording_mbid anchors a single-track request raised from a
-    -- track-level want.
-    recording_mbid     TEXT,
-    artist             TEXT    NOT NULL DEFAULT '',
-    album              TEXT    NOT NULL DEFAULT '',
-    query              TEXT    NOT NULL DEFAULT '',
-    expected           TEXT    NOT NULL DEFAULT '[]',
-    state              TEXT    NOT NULL DEFAULT 'searching'
-        CHECK(state IN ('searching', 'found', 'queued', 'grabbing',
-                        'verifying', 'tagging', 'importing',
-                        'complete', 'cancelled', 'failed')),
-    error              TEXT    NOT NULL DEFAULT '',
-    created_at         DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
-    updated_at         DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
-    FOREIGN KEY(library_id) REFERENCES libraries(id) ON DELETE CASCADE
+    id         INTEGER PRIMARY KEY AUTOINCREMENT,
+    mbid       TEXT    NOT NULL,
+    entity     TEXT    NOT NULL
+        CHECK(entity IN ('artist', 'release-group', 'release', 'recording')),
+    library_id INTEGER NOT NULL,
+
+    -- Display text, cached so the list renders without touching the
+    -- explore index.  Neither is authoritative; the MBID is.
+    artist     TEXT    NOT NULL DEFAULT '',
+    title      TEXT    NOT NULL DEFAULT '',
+
+    scope      TEXT    NOT NULL DEFAULT 'future'
+        CHECK(scope IN ('future', 'all')),
+
+    -- secondary controls whether an artist request's expansion includes
+    -- compilations, live albums and remixes.  Off by default: someone
+    -- subscribing to an artist wants the albums, not six versions of
+    -- the same greatest-hits package.
+    secondary  INTEGER NOT NULL DEFAULT 0,
+
+    state      TEXT    NOT NULL DEFAULT 'wanted'
+        CHECK(state IN ('wanted', 'satisfied', 'paused')),
+
+    -- parent_id links a request the reconciler derived from an artist
+    -- request.  Deleting the artist takes its derived children with
+    -- it, but children the user pinned themselves have no parent and
+    -- stay.
+    parent_id  INTEGER,
+
+    -- Retry bookkeeping.  attempts drives the backoff; last_error is
+    -- the most recent reason it did not work out, which for a request
+    -- is information rather than a failure.
+    attempts     INTEGER  NOT NULL DEFAULT 0,
+    last_error   TEXT     NOT NULL DEFAULT '',
+    last_tried_at DATETIME,
+    next_try_at  DATETIME,
+
+    -- external_ids maps provider row ID to that provider's own
+    -- identifier for this request, for clients that keep their own
+    -- persistent list (a Lidarr artist ID).  JSON object.
+    external_ids TEXT    NOT NULL DEFAULT '{}',
+
+    created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updated_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
+
+    -- One request per thing per library.  Asking twice is not two
+    -- requests, and this is what lets an artist expansion re-run every
+    -- reconcile without accumulating duplicates.
+    UNIQUE(mbid, library_id),
+
+    FOREIGN KEY(library_id) REFERENCES libraries(id) ON DELETE CASCADE,
+    FOREIGN KEY(parent_id) REFERENCES download_requests(id) ON DELETE CASCADE
 );
 
-CREATE INDEX IF NOT EXISTS idx_download_requests_created
-    ON download_requests(created_at DESC);
-
-CREATE INDEX IF NOT EXISTS idx_download_requests_state
-    ON download_requests(state);
+-- idx_download_requests_{due,entity,parent} are deliberately NOT
+-- declared here. This table name is reused from the old one-shot
+-- attempt table (also called download_requests before the Want/Request
+-- rename), so on an existing database this CREATE TABLE is a no-op
+-- against a table that, at schema-pass time, is still shaped like the
+-- OLD attempts table and lacks these columns entirely — an inline
+-- CREATE INDEX here would fail outright rather than just no-op. See
+-- migrateDownloadRename/ensureDownloadIndexes in
+-- backend/database/download_rename_migration.go, which create these
+-- once the rename has actually happened (or immediately, on a fresh
+-- database where the columns exist from the start).