WL#2754: enable event_api to receive and post write errors from Mysqld

If the Mysqld runs out of disk space or just has isses writting to the bin log
or error log, then it should pass that information back to the event_api so that
it can be recorded into the cluster log.

We had an issue today when one of the three mysqld processes ran out of disk
space. Looking at the cluster log, it was not apparent right off that was the
issue. After Tomas and I looked around for a while tomas notice that the error
messages where comming from a different node then what the master node was. Once
he looked there it was just a matter of "time" before he found it out of space.

Since the cluster logs and files where on a different drive, an error message of
not being able to write or out od disk space would have sped the process up.

Below is the error message that pointed us to Node 17. The GCIs on that system
were out of wack because it could not write to the bin log. This also caused
transaction errors for the TPC-B scripts that were running.

2005-07-23 17:13:31 [MgmSrvr] INFO     -- Node 17: Event buffer status:
used=17MB(100%) alloc=17MB(0%) max=0B apply_gci=50803 latest_gci=55503