Your first equation is correct in that it takes 316 volts to produce 2KW at 50 ohms. That's the only relevant number we need to concern ourselves with going forward.
Your next equation is where the escape occurs. 316 volts into an 18k load produces 5.55 watts of total heat, not 5.55 watts at each resistor. Per my earlier post I have 9 resistors in a 3x3 parallel-series arrangement, so each resistor is sharing an equal part of the load. 5.55 divided by 9 is .616, so each resistor is dealing with .616 watts of heat load at 2KW. My little AL-80B is only capable of half that power so each resistor is only subject to .308 watts max heat load. Those are all large 2 watt resistors, so the hardest my amp is able to push them is effectively less than 1/6 of their rating.
Regardless, as a sanity check after fab I subjected the coupler to several hours of full-power operation using various transmission modes (SSB, AM, digital) on multiple bands, and never once did the resistor pack feel anything other than stone cold to my touch.
That coupler is way overbuilt.