Avoid full-table scan in /recordings/unavailable

The query filtered on (end_time >= after AND start_time <= before). With
before defaulting to now, start_time <= before matches every row, so SQLite
range-scans the entire recordings table for any window. Add a start_time lower
bound (segments are short, so a recording overlapping the window starts within
an hour of it). Also fix a 500 on cameras=all, where cameras was reassigned to a
list and later .split().

frigate/api/record.py

@@ -274,7 +274,7 @@ async def no_recordings(
             return JSONResponse(content=[])
         cameras = ",".join(filtered)
     else:
-        cameras = allowed_cameras
+        cameras = "all"
 
     before = params.before or datetime.datetime.now().timestamp()
     after = (
@@ -283,7 +283,11 @@ async def no_recordings(
     )
     scale = params.scale
 
-    clauses = [(Recordings.end_time >= after) & (Recordings.start_time <= before)]
+    clauses = [
+        (Recordings.start_time >= after - 3600)
+        & (Recordings.start_time <= before)
+        & (Recordings.end_time >= after)
+    ]
     if cameras != "all":
         camera_list = cameras.split(",")
         clauses.append((Recordings.camera << camera_list))