I'm not quite sure what you mean by "Linear". If you mean reading properly at all levels of energy ... that is a valid concern.

I haven't calibrated photographic "Exposure Meters", but I have calibrated something one or two magnitudes more accurate and sensitive - Cascade Photometers. That calibration required an Optical Bench (or similar) and a Standard Lamp ... we had Intralaboratory Standard Lamps, rotating to N.B.S. They would be powered by a constant AMPERAGE power source (not voltage ... the light output is three times more sensitive to amperage than voltage). Those photomultipliers read in Watt-seconds (or milli-watt seconds) and reacted to energy, in the form of light. An "exposure meter" does the same thing, and mathematically "massages" the readings to be useful in photography.
Someone tried to use one of the original Honeywell 1/21 exposure meters in our optical work ... I wrung that puppy out, and, after an involved conversation with Honeywell, determined that it was within manufacturer's specifications; +/- one half "stop" ... altogether too coarse for what we were doing, but fine for photography.

That exercise resulted in valuable information ... I cringe whenever I hear a photographer claim that he ALWAYS exposes his film, transparency or negative, "within a tenth of a stop". Yeah -- right. He would have absolutely no way of knowing if that was true. So far, the best accuracy claim I've heard of was +/- 1/3 "stop" and I'm wondering about the truth of that, using a handheld meter.