I would rather see a single diode capable of the load. Diodes don't always share well, even with a small series resistor, since small differences in forward voltage will upset the current sharing. Also, if one fails, it could quickly become a string of firecrackers as the load gets bigger as each successive diode fails.
A single diode capable of this would be in a stud mount package ( 1N3212 for example ) would need heatsinking, and obviously all of this needs to be isolated electrically from the hotplates metal which is surely grounded.
The diode is probably about 5 dollars, you could make a heatsink from some alluminum, or buy one for maybe another 5 or 10 dollars. then you need to build it all, and do it safely. If you need to pay for shipping of any of the parts, it adds to the cost.
Add it all up and maybe selling the hot plate or keeping it as a hot plate, and spending the money on a used lab plate maybe more cost effective.
Plan#2
As suggested, I would try to modify the bimetal switch, and maybe put a metal plate, say 1/4" thick on the burner to add thermal mass. This way even if the burner cycles really slowly, the metal plate will have stored thermal energy to "smooth out" the pulsing in the heat. This could be done pretty cheap.