I couldn't follow what that article was saying very well. He didn't address what would make such high currents in ground, only some current should flow.
"There is no way to get 100 amps through a 120 volt cord-connected product on a 20 amp branch circuit.
The board did burn. It did take 100 amps to bum the board. Those 100 amps had to come from somewhere."
The 100 amps had to come from somewhere. Through the 20 amp branch breaker? Through the 20 amp branch circuit neutral?
If not one of those, could be generated by a step-down transformer. 1/5 the voltage, 5x the amps to get 100 amps.
"Have you ever measured the potential difference between the neutral and the ground? You probably measured a couple of volts.
To get a few volts potential difference, there must be a few amps of current passing through a resistance, somewhere."
Normally, if you measure neutral to ground, there is a small voltage. If you connect them, there is a small current.
When a 200A single-phase load (24kW 120V) is connected in your house, 200A flows through L1, 200A through L2. This would be a branch circuit hopefully wired in 2/0.
If your 20A branch circuit was also wired to same sub-panel, its neutral would see elevated voltage. Assume the circuit's 12 awg wire was very short, low in resistance compared to the 2/0 wire from sub panel back to main panel. Some of the 200A neutral current flows through 12 awg neutral, to your PCB, through the redundant ground, back through circuit's ground wire to sub panel. Now, it can split between ground (normally smaller than 2/0) and neutral (2/0) back to main panel. Less than 100A flows through your circuit, because ground wire is higher resistance than neutral wire.
You have to try pretty hard to get excessive "objectionable" current in ground. Bad neutral connection on the 200A load, running hot, could do it.
Something is wrong with the wiring in forum member's system, but I doubt multiple neutral paths is the problem. Would help if we knew what burned out.
Isolating the two inverters so they only share incoming neutral and ground for testing (and measurement of voltages between them), then connecting data cable for paralleling and retesting, could shows something. Keep outputs isolated while looking for voltage between wires you plan to connect.