So far as I can tell, Petra called that a ring formed from 10 days worth of seismicity, with 10 earthquakes, and offered a 10-day prediction. This is for an area where the mainshock hit in an area with a long-term average rate of somewhat fewer earthquakes than the surrounding region, according to my impression of the figures in the link I gave.

More than a month later, an earthquake hits, in Tonga, which has the highest rate of earthquakes in the world. The prediction is for an M6.8+, not such a rare event. Noone has checked whether the donut lasted until the mainshock or not, nor compared the pattern made by the ten earthquakes with the background seismicity.

Again, one prediction, not filled according to the specified time window, with no chance of statistical analysis, contradicting existing studies with much more data showing Mogi donuts are not a general feature of regions about to have a large earthquake.

Can't say I'm convinced.