The following shows the progression to re-produce this. Is this a bug or some expected behavior when not using the -z option?

MariaDB Server:
MariaDB [repl_examples]> create table z_error(a int(11), b int(11));
Query OK, 0 rows affected (0.006 sec)

MariaDB [repl_examples]> insert into z_error values(1,2),(3,4),(5,6),(7,8),(9,10);

MaxScale:
[root@usfit-scmsdev2 maxscale]# mxs_adapter -c /etc/ColumnStore.xml -u cdcuser -p cdc -h localhost -P 4001 -r 10000 -d -n -a -f tables.lst
2019-08-27 09:51:41 [main] Started thread 0x12522a0
2019-08-27 09:51:41 [main] Started 1 threads
2019-08-27 09:51:41 [repl_examples.z_error] Requesting data for table: repl_examples.z_error
2019-08-27 09:51:41 [repl_examples.z_error] Creating table: CREATE TABLE repl_examples.z_error (a int(11), b int(11)) ENGINE=ColumnStore;
2019-08-27 09:51:43 [repl_examples.z_error] Created table `repl_examples`.`z_error`
2019-08-27 09:51:43 [repl_examples.z_error] Requesting data for table: repl_examples.z_error
2019-08-27 09:51:43 [repl_examples.z_error] a: 1
2019-08-27 09:51:43 [repl_examples.z_error] b: 2
2019-08-27 09:51:43 [repl_examples.z_error] a: 3
2019-08-27 09:51:43 [repl_examples.z_error] b: 4
2019-08-27 09:51:43 [repl_examples.z_error] a: 5
2019-08-27 09:51:43 [repl_examples.z_error] b: 6
2019-08-27 09:51:43 [repl_examples.z_error] a: 7
2019-08-27 09:51:43 [repl_examples.z_error] b: 8
2019-08-27 09:51:43 [repl_examples.z_error] a: 9
2019-08-27 09:51:43 [repl_examples.z_error] b: 10
2019-08-27 09:51:53 [repl_examples.z_error] Flushing batch
2019-08-27 09:51:54 [repl_examples.z_error] 5 rows, 0 transactions inserted over 10.3971 seconds. GTID = 0-1-100:5

Columnstore:
MariaDB [repl_examples]> select * from z_error;
----------+

----------+

----------+

MariaDB Server:
MariaDB [repl_examples]> delete from z_error where a=1;
Query OK, 1 row affected (0.002 sec)

Maxscale:
2019-08-27 09:53:03 [repl_examples.z_error] a: 1
2019-08-27 09:53:03 [repl_examples.z_error] b: 2
2019-08-27 09:53:04 [repl_examples.z_error] 1 rows, 0 transactions inserted over 0.239964 seconds. GTID = 0-1-101:1
2019-08-27 09:53:14 [repl_examples.z_error] Read timeout with empty batch, closing it
2019-08-27 09:53:25 [repl_examples.z_error] Read timeout
2019-08-27 09:53:35 [repl_examples.z_error] Read timeout

Columnstore:
MariaDB [repl_examples]> select * from z_error;
----------+

----------+

----------+
6 rows in set (0.014 sec)

Server:
MariaDB [repl_examples]> select * from z_error;
----------+

----------+

----------+

As you can see.. the deleted row from InnoDB was not deletd (because the option of -z was not specified), but it was re-inserted then into Columnstore.