The issue was that the test did not take into account that the IO thread
could have been in COMMAND=Connecting state, which happens before the
COMMANMD=Slave_IO state.

The test is a bit fragile as it depends on the COMMAND state to be
syncronised with the Slave_IO_State, which is not the case.

I added a new proc state and some more information to the error
output to be able to diagnose future failures more easily.