24.9 GB of index storage has not been read once during the observation window. Every one of them still has to be updated on every write.

Business Impact: Unused indexes tax every insert, update and delete, and add work to every cleanup pass. The cost is write latency, not just disk.

Postgres does not erase updated or deleted rows immediately — it marks them dead and cleans up later. 92.4% of your busiest table is dead rows.

Business Impact: Dead rows are not just wasted disk: every full scan still walks over them, so queries stay slow until cleanup catches up.

Automatic cleanup can only remove rows older than the oldest thing still looking at them. 2 blocker(s) is holding that line — so cleanup is frozen for the entire database, not just one table.

Business Impact: This is why the bloat above cannot be fixed by "just running a vacuum". Every table grows at once until the blocker is cleared.