Percona fails to properly stop replication thread if super_read_only=1, autocommit=0 and >=5.6.29

[Arturas Moskvinas]

Percona 5.6.29-76.2-1 fails to properly stop replication if super_read_only mode is on and session is in autocommit=0 mode. Actually it is stopping thread but also returns error:
```
ERROR 1290 (HY000): The MySQL server is running with the --read-only (super) option so it cannot execute this statement
```

How to reproduce:
1. Setup slave replication
2. Execute `echo 'set global super_read_only=1; set autocommit=0; stop slave sql_thread' | mysql`

Result:
```
ERROR 1290 (HY000) at line 1: The MySQL server is running with the --read-only (super) option so it cannot execute this statement
```
And SQL_THREAD is actually - stopped.

Expect:
No error+stop thread or previous error and sql_thread still running...

[Arturas Moskvinas (arturas-w)]

Some essential settings we use:
```
relay_log_info_repository = TABLE
default-storage-engine = innodb
```

[Arturas Moskvinas (arturas-w)]

Not only `stop slave` is affected, you cannot `change master` also when autocommit=0

[Jericho Rivera (jericho-rivera)]

This is easily reproducible on latest PS 5.6 and 5.7, confirmed that I had to add relay-log-info-repository=TABLE and master-info-repository=TABLE to reproduce.

slave1 [localhost] {root} ((none)) > set global super_read_only=1; set autocommit=0; stop slave sql_thread;
Query OK, 0 rows affected (0.00 sec)

Query OK, 0 rows affected (0.00 sec)

ERROR 1290 (HY000): The MySQL server is running with the --super-read-only option so it cannot execute this statement

[Jericho Rivera (jericho-rivera)]

@Arturas, were you able to test if this affects upstream MySQL as well?

[Arturas Moskvinas (arturas-w)]

Upstream MySQL 5.6 is not supporting `super_read_only` option, it is only available in 5.7, we do not have any production databases running 5.7 at the moment.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-1716