I can understand the edge case RCD risk.
Very limited set of circumstances where say 30mA of fault current (say you touch a faulty appliance when standing on carpet in your house - high impedance, therefore low current shock), would have normally tripped the RCD but now it doesn't because some of that is hidden from it by the solar.
Before RCDs we didn't have this protection at all. I guess a good way to describe it is we've gone from no RCDs (100% risk) to full RCDs (1% risk) and with plug in solar we might regress ever so slightly to 2% risk due to the edge cases. Obviously not real numbers just to illustrate the point.
I think that article is probably intentionally misleading - the risk of plug in solar causing direct overheating of cabling is not the concern at all.
The edge case RCD risk, yes I can understand that one.
Edit - just to add, a 'bidirectional' RCD doesn't resolve this issue either. That just means it works when the solar is pushing current back through to the grid. It doesn't change the risk of failure to trip due to sharing of fault current.
Edit 2- just to add as well to demonstrate how edge case this is - there are probably still many thousands of houses in the UK that don't have RCDs at all. My parents house still has fuses! This RCD issue on plug in solar makes absolutely no difference if you don't have RCDs in the first place.