truncate after select may cause distributed deadlock in mx nodes

Observed a distributed deadlock when testing a recent work on allowing truncate on MX nodes.

Verified that the deadlock does not occur on single node (not distributed) configuration.

- create 2 node cluster with both workers as mx nodes
- create a distributed table table_1

- connect to worker 2 (the one added to mx later) (connection 1)
- start a transaction
- issue select * from table_1

- connect to worker 1 (connection 2)
- start a transaction
- issue truncate table table_1 --> will block

- switch to connection 1
- issue truncate table table_1 --> will block

After a while our distribution deadlock detection kicks in an cancel one of the transactions.