This shows a typical 'show processlist' when using replication

| 6 | system user |...| Slave_IO     | 177806 | Waiting for master to send event

| 8 | system user |...| Slave_worker |      0 | Waiting for prior transaction to commit

| 9 | system user |...| Slave_worker |      0 | closing tables

| 7 | system user |...| Slave_SQL    |    259 | Waiting for room in worker thread event queue

There is a couple of problems with the above result:

• Slave_IO shows the time since SLAVE START, not the time it has been waiting for the master
• Slave_worker's often have a state 'closing tables', because closing_tables() does not
  revert back to previous state when it is done. This gives a wrong impression of what the Slave_worker is doing.
• Slave_SQL has a wrong (random time). A better time would be how long it has been waiting for the event.

Slave_worker threads also does not show if they are waiting for commit or rollback.

The purpose of this task is to correct the times, fix 'closing tables' and add stages for commit and rollback.