bug(indexer): Flaky `test_optimistic_tables_pruning`

[tomxey]

Sometimes on CI the test fails with following error:

failures:

---- write_api::test_optimistic_tables_pruning stdout ----

thread 'tokio-runtime-worker' panicked at crates/iota-indexer/src/store/pg_indexer_store.rs:1720:9:
you are calling a blocking DB operation directly on an async thread. Please use IndexerReader::spawn_blocking instead to move the operation to a blocking thread
stack backtrace:

Which traces to the call to:

fn get_watermarks(&self) -> Result<(Vec<StoredWatermark>, i64), IndexerError> {
        // read_only transaction, otherwise this will block and get blocked by write
        // transactions to the same table.
        run_query_with_retry!(
            &self.blocking_cp,
            |conn| {
                ...
            },
            PG_DB_COMMIT_SLEEP_DURATION
        )
    }

But the function seems to be wrapped in a blocking task:

    async fn get_watermarks(&self) -> Result<(Vec<StoredWatermark>, i64), IndexerError> {
        self.execute_in_blocking_worker(move |this| this.get_watermarks())
            .await
    }

To be investigated what is the exact cause.