it seems perfectly plausible to me that the tower could visual separate the aircraft at night with the go around off 34 passing behind the 27 go around. Without more timings or an actual radar paint that is speculation
Surely with limited ability to adjust speed, whether one aircraft passes behind is set by the spacing on final, i.e. if they are at 3.5 and 4 nm the spacing will be near enough 0.5 mile behind. Maybe that is a masterpiece if planning and timing, or maybe it is just how the cards fell on the day.
The assumption at that stage is that you don't have both go around (once in 175 years). There may be very little time to do anything once they do go around. Do ATC really visually judge crossing aircraft that finely at night (or even during the day?)
The report also suggests that turning one aircraft (for wake turbulence avoidance) below minimum vector altitude was not the right thing to do, which implies there is little you can do by turning aircraft either. Not that a turn would make much difference for aircraft crossing at near enough 90 degrees.
If speed and direction are fixed, altitude is dictated by the parameters of the missed approach, exactly what tools do ATC have to adjust separation in this situation - visually or otherwise?
OJTI and ATSB also thought it was not an issue
Some may suggest that Airservices are trying to play it down, and ATSB are prepared to assist.