How close to high power RF do you really want to get?
Personally I'd insert a 20dB directional coupler (appropriate power and frequency), and measure the output of that.
100watts is hardly high power!
The voltage needed for that power level in 50ohms is 70.71 volts RMS.
The directional coupler is an elegant solution, but has the disadvantage that you are unlikely to get one that works satisfactorily from HF to UHF.
An attenuator is a reasonable idea, but the OP needs to remember that is the device where most of the power is dissipated, with the test load only handling "what's left".
A power attenuator, (what we used to call a "through load") will work well, as will a test load with a tapping or built up from individual resistors to do the same thing.