feat(backend): duplicate pairs, dismissals, and merge resolution

Adds the duplicate-detection backend on top of perceptual hashing:

- Two tables (edited into the original migrations): data.duplicate_pairs holds
  precomputed near-duplicate candidates (rebuilt wholesale by the rescan), and
  data.duplicate_dismissals is a global "not a duplicate" overlay that survives
  rescans. New audit actions file_merge / duplicate_dismiss.
- DuplicateService:
  - Rescan builds every pair within DUPLICATE_HASH_THRESHOLD via a BK-tree over
    the perceptual hashes and replaces the pairs table. This is the only thing
    that populates pairs, so GET never compares all-vs-all (scales to 110k+).
  - Clusters reads the precomputed pairs (ACL-filtered, non-trashed, non-
    dismissed), groups them into connected components via union-find, and
    paginates whole clusters.
  - Resolve merges a pair field-by-field: each scalar from keep or discard,
    metadata keep/discard/shallow-merge, tags/pools keep or union; then trashes
    the discarded file. Enforces edit ACL on both.
  - Dismiss records a canonical pair (view ACL on both).
- Endpoints under /files: GET /files/duplicates, POST /files/duplicates/dismiss,
  POST /files/duplicates/resolve (registered before /:id to avoid collision).
  Plain delete reuses /files/bulk/delete.
- Repo support: ListMissingPHash, ListAllPHashes, CopyPoolMemberships, plus the
  DuplicatePairRepo (ReplaceAll via COPY, ListVisible) and DismissalRepo.

Unit tests cover the BK-tree pairing, union-find clustering, metadata merge and
field validation; an integration test covers rescan -> list -> merge -> dismiss
(including that a dismissal survives a re-rescan).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

backend/migrations/003_data_tables.sql

@@ -92,6 +92,31 @@ CREATE TABLE data.file_pool (
     PRIMARY KEY (file_id, pool_id)
 );
 
+-- Precomputed near-duplicate candidates (phash Hamming distance <= threshold),
+-- (re)built in full by the dedup rescan. Stored once per unordered pair with a
+-- canonical file_a < file_b ordering so a pair is never duplicated as (a,b)/(b,a).
+CREATE TABLE data.duplicate_pairs (
+    file_a   uuid     NOT NULL REFERENCES data.files(id) ON UPDATE CASCADE ON DELETE CASCADE,
+    file_b   uuid     NOT NULL REFERENCES data.files(id) ON UPDATE CASCADE ON DELETE CASCADE,
+    distance smallint NOT NULL,
+
+    CONSTRAINT chk__duplicate_pairs__order CHECK (file_a < file_b),
+    PRIMARY KEY (file_a, file_b)
+);
+
+-- "Not a duplicate" decisions: a global overlay that hides a candidate pair from
+-- the duplicates view. Survives rescans (the pair may be re-found but stays
+-- hidden). Same canonical file_a < file_b ordering as data.duplicate_pairs.
+CREATE TABLE data.duplicate_dismissals (
+    file_a       uuid        NOT NULL REFERENCES data.files(id) ON UPDATE CASCADE ON DELETE CASCADE,
+    file_b       uuid        NOT NULL REFERENCES data.files(id) ON UPDATE CASCADE ON DELETE CASCADE,
+    dismissed_by smallint    NOT NULL REFERENCES core.users(id) ON UPDATE CASCADE ON DELETE RESTRICT,
+    dismissed_at timestamptz NOT NULL DEFAULT clock_timestamp(),
+
+    CONSTRAINT chk__duplicate_dismissals__order CHECK (file_a < file_b),
+    PRIMARY KEY (file_a, file_b)
+);
+
 COMMENT ON TABLE  data.categories IS 'Logical grouping of tags';
 COMMENT ON TABLE  data.tags       IS 'File labels/tags';
 COMMENT ON TABLE  data.tag_rules  IS 'Auto-tagging rules: when when_tag is assigned, then_tag follows';
@@ -99,6 +124,8 @@ COMMENT ON TABLE  data.files      IS 'Managed files; actual content stored on di
 COMMENT ON TABLE  data.file_tag   IS 'Many-to-many: files <-> tags';
 COMMENT ON TABLE  data.pools      IS 'Ordered collections of files';
 COMMENT ON TABLE  data.file_pool  IS 'Many-to-many: files <-> pools, with ordering';
+COMMENT ON TABLE  data.duplicate_pairs       IS 'Precomputed near-duplicate candidate pairs (perceptual-hash distance)';
+COMMENT ON TABLE  data.duplicate_dismissals  IS 'Pairs marked "not a duplicate"; hidden from the duplicates view';
 
 COMMENT ON COLUMN data.files.original_name    IS 'Original filename at upload time';
 COMMENT ON COLUMN data.files.content_datetime IS 'Content datetime (e.g. when photo was taken); falls back to EXIF DateTimeOriginal';
@@ -110,6 +137,8 @@ COMMENT ON COLUMN data.file_pool.position     IS 'Manual ordering within pool; u
 
 -- +goose Down
 
+DROP TABLE IF EXISTS data.duplicate_dismissals;
+DROP TABLE IF EXISTS data.duplicate_pairs;
 DROP TABLE IF EXISTS data.file_pool;
 DROP TABLE IF EXISTS data.pools;
 DROP TABLE IF EXISTS data.file_tag;

backend/migrations/006_indexes.sql

@@ -26,6 +26,12 @@ CREATE INDEX idx__files__needs_review      ON data.files USING btree (id) WHERE
 CREATE INDEX idx__file_tag__tag_id  ON data.file_tag USING hash (tag_id);
 CREATE INDEX idx__file_tag__file_id ON data.file_tag USING hash (file_id);
 
+-- data.duplicate_pairs / data.duplicate_dismissals
+-- The composite primary keys cover lookups on file_a; these add the file_b side
+-- (used by the ON DELETE CASCADE and by the visibility join on the second file).
+CREATE INDEX idx__duplicate_pairs__file_b      ON data.duplicate_pairs      USING hash (file_b);
+CREATE INDEX idx__duplicate_dismissals__file_b ON data.duplicate_dismissals USING hash (file_b);
+
 -- data.pools
 CREATE INDEX idx__pools__creator_id ON data.pools USING hash (creator_id);
 
@@ -70,6 +76,8 @@ DROP INDEX IF EXISTS activity.idx__sessions__token_hash;
 DROP INDEX IF EXISTS activity.idx__sessions__user_id;
 DROP INDEX IF EXISTS acl.idx__acl__user;
 DROP INDEX IF EXISTS acl.idx__acl__object;
+DROP INDEX IF EXISTS data.idx__duplicate_dismissals__file_b;
+DROP INDEX IF EXISTS data.idx__duplicate_pairs__file_b;
 DROP INDEX IF EXISTS data.idx__file_pool__file_id;
 DROP INDEX IF EXISTS data.idx__file_pool__pool_id;
 DROP INDEX IF EXISTS data.idx__pools__creator_id;

backend/migrations/007_seed_data.sql

@@ -21,6 +21,7 @@ INSERT INTO activity.action_types (name) VALUES
     -- Files
     ('file_create'), ('file_edit'), ('file_delete'), ('file_restore'),
     ('file_permanent_delete'), ('file_replace'), ('file_review'),
+    ('file_merge'), ('duplicate_dismiss'),
     -- Tags
     ('tag_create'), ('tag_edit'), ('tag_delete'),
     -- Categories