To be clear regarding the important point I think that Chue X might be making: a relay has excellent isolation between the line and load terminals while a triac does not. For example, the BT136-600 datatsheet shows that this 4A triac a max leakage of 0.5 mA. That's a triac that would be suitable for a regular wall dimmer. Unless the dimmer includes a mechanical switch, then you would measure 120 VAC on the load side when the triac is off if there is no connected load. If there is a load connected you will measure a much lower voltage which would be equal to the leakage current times the load resistance.
As a rule of thumb, you would expect that a higher power triac capable of 4 KW would have higher leakage current due to its much larger active area. That would create a substantial shock hazard in the oven when the element is burned out or needs to be removed for service. There would be 230 VAC with significant current capability exposed on the heating element connections. Using a relay insures that the element is safely isolated from the line when the oven is off.
Regarding opto-isolated triacs: that is referring to isolation between the line/load connections and the control connections. That is needed to prevent dangerous voltages and currents flowing back through the control input signal to electronics that are controlling it. A good overview of optocouplers, including opto-isolated triacs is available here Optocoupler tutorial. Opto-isolated triacs still have substantial leakage current and are frequently not suited for controlling certain loads. This is also the sort of leakage that a relay provides between its coil and load, as mentioned by Olin's answer.