Xavier60, regarding the IGBT failures - I wonder if the magnetron is ageing and presenting a different load as it gets older. ie., its equivalent resistance changes and alters the damping of the tank circuit perhaps increasing the peak voltage and current. Also, when the inverter goes off at the end of a cycle, perhaps any stored energy has to be dumped as a safety feature to prevent emission with the door open and maybe the inverter can't handle the increased energy.
I have a couple of older microwave ovens. One had a shorted triac that allowed the magnetron to continue to run after the time expired. The door had to be opened to completely stop it. If the normally closed door switch stuck closed at the at same time, it could have caused some excitement.