"Replica(s) failed to execute read" on simple select on stress-created table with >1 nodes

This looks similar to CASSANDRA-9825, though that was marked as resolved.

To reproduce, create a multi-node cluster and insert values using stress. Then, try to run a SELECT statement on the keyspace1.standard1 table created by stress.

Running the select through the Python driver, I consistently get the result

cassandra.ReadFailure: code=1300 [Replica(s) failed to execute read] message="Operation failed - received 0 responses and 1 failures" info={'failures': 1, 'received_responses': 0, 'required_responses': 1, 'consistency': 'ONE'}

I've attached a script that reproduces it via the ccm CLI. As the script also demonstrates, the error doesn't show under 2.2, or when there is only 1 node in the cluster.

In the repro script, I've also included a commented-out insertion via cqlsh. If you uncomment that and comment out the stress line, the script demonstrates that the error doesn't occur when the table was created/inserted into via cqlsh.