I wouldn’t want the load current flowing through anything it shouldn’t. Maybe I’m too cautious, but it’s a binary rule that I’ve always followed when the load current is enough to cause harm to anything, or release the magic smoke inside the controller.

To your point, I have had the high voltage side of a high-efficiency furnace (e.g., inducer fan) fail and take out the entire string of low voltage components (thermostat, humidifier controls) connected to the furnace through an unfortunate coupling on the primary control board.

I wouldn't want that, either.

But a relay is already an isolator: There's no path between the switched contacts and the coil of the solenoid that switches those contacts. The coil doesn't know or care how much current or voltage the contacts are switching: It's just a coil that moves a little bit of metal using electromagnetism.

And, sure: It takes more power to drive a relay's coil than it does to drive an LED inside of an optocoupler, but at some point that coil still needs to be driven by something that can accomplish that small amount of work.

Adding an optocoupler just seems like adding extra parts to a circuit that would be simpler without.

KISS?