Clean up the DistributedQueryExecution table

[zhuoyuan-liu]

It looks like for each of the nodes, we relay on the function NotYetExecuted to check if the distributed query is executed.

When each node finishes the distributed query, it will write a record in this table. However, I didn't find any code that remove these records, and it would become an issue when we run the distributed query frequently. I believe it's a good idea to forget or remove these records when the corresponding query is completed or failed.