In an ideal world, I'd want a continuous plot from 0° to 90°. In practice without a automated system for doing this, it would take impractically long to undertake.
I wasn't so much trying to get an idea of how well it was visible from rear approaching traffic, but how wide the overall spread was (and significant off angle spread would be useful at junctions, on curved roads). Like front lights, some rear lights have very bright but very narrow beams, whereas others spread the light over a wide beam.
It would be easier to measure some characteristics of the width by going significantly off bore. If you just go a little bit off bore, the light level may be 90% or 88% down, the difference between which would be fiddly to measure. If you went significantly off bore, say 30° or 45°, the drop would be more significant, as would be the differences, so noise in the measurement would be less important.