Restored database fails integrity checks

[kalafut]

I'm using version v0.3.13 in a mostly vanilla fashion: single server replicating to S3. The only modifications have been to the retention setup:

retention: 168h
snapshot-interval: 24h

Things had been working fine, including test restores. Even now replication is proceeding as normal. But about a week ago, I did a test restore and saw some database integrity errors. When I ran pragma integrity_check I saw stuff like:

*** in database main ***
Tree 88 page 88 cell 46: Rowid 5452 out of order
Tree 88 page 35: btreeInitPage() returns error code 11
Tree 88 page 88 cell 43: Rowid 5436 out of order
Tree 88 page 88 cell 39: Rowid 5402 out of order
Tree 88 page 88 cell 38: Rowid 5393 out of order
Tree 88 page 88 cell 37: Rowid 5384 out of order
...
Tree 88 page 88 cell 7: Rowid 5233 out of order
Tree 88 page 91 cell 1: 2nd reference to page 35
row 22 missing from index idx_objects_user_id_modified_at
row 34 missing from index sqlite_autoindex_objects_1
row 35 missing from index sqlite_autoindex_objects_1
row 36 missing from index sqlite_autoindex_objects_1
...
row 49 missing from index idx_objects_user_id_modified_at
row 49 missing from index sqlite_autoindex_objects_1
row 50 missing from index idx_objects_user_id_modified_at
row 50 missing from index sqlite_autoindex_objects_1

I restarted the Litestream service and just let things run, thinking maybe it was a glitchy WAL and would be flushed out (still not great, but...). A week later, a freshly restored file still exhibits the same issues. When I ssh into the server and run pragma integrity_check on the actual database, it always returns OK.

I'm sure if I just wiped/vacuumed/restarted everything, thing would probably again run fine, but that's not answering any questions. Are there any diagnostics you'd like me to run on this bad state?