feat(arr): one sidecar per language, in the schema

DESIGN.md §15 as amended: a language is satisfied by exactly one
sidecar, and no filename segment distinguishes forced from plain from
SDH. A unique index over sidecar rows says so; embedded rows keep their
own key, since several tracks for one language can legitimately coexist
inside a video.

Existing databases may hold a duplicate from a manual grab that beat the
API's path check, so the migration resolves them rather than failing: a
real subtitle beats a machine translation, and of two of the same kind
the newest wins. The files stay on disk for the manual delete to clean
up.

`record_file` no longer swallows every conflict — only the two that mean
"arr already knows this file".

.sqlx/query-3d48b78768bf8dc2e337d0935889a5654089f38bb7ee2a15b7e19f4ed5540262.json

@@ -0,0 +1,26 @@
+{
+  "db_name": "SQLite",
+  "query": "INSERT INTO subtitle_files\n               (media_file_id, language, origin, provider, candidate_id, engine,\n                forced, sdh, synced, sync_rejected, path)\n           VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)\n           ON CONFLICT (path) DO NOTHING\n           ON CONFLICT (media_file_id, language, forced, sdh)\n               WHERE origin = 'embedded' DO NOTHING\n           RETURNING id AS \"id!: i64\"",
+  "describe": {
+    "columns": [
+      {
+        "name": "id!: i64",
+        "ordinal": 0,
+        "type_info": "Integer",
+        "origin": {
+          "Table": {
+            "table": "subtitle_files",
+            "name": "id"
+          }
+        }
+      }
+    ],
+    "parameters": {
+      "Right": 11
+    },
+    "nullable": [
+      null
+    ]
+  },
+  "hash": "3d48b78768bf8dc2e337d0935889a5654089f38bb7ee2a15b7e19f4ed5540262"
+}