What if you're rinsing it though? As in adding in say 2-3x the volume of cooled boiled water as you have slurry, swirling it up, and letting it settle out for a few minutes then pitching the liquid. It would be better to set it at a low non-yeast percentage and low yeast concentration, right?
Whatever methods gives you the most accurate count, and ime it's using the settled yeast volume. If you took 100mL of settled yeast and mixed it with 300mL water I'm not sure it would look all that different than 200mL yeast + 200mL water once you shook it all up.
This is a question that I see all the time on homebrew forums and nobody seems to have a definitive answer for yeast count per unit volume of settled yeast. Or at least the ones that claim to don't agree. We have yeast count but then we also have viability and non-yeast percentage numbers and everybody seems to be guessing at all three. It seems the yeast count of settled yeast part should be straightforward enough. Guessing at two variables is better than guessing at three.
There is a picture somewhere from Wyeast or WhiteLabs showing vials with different yeast densities in them; of course I can't find it when I am looking for it.
That being said, I think it still really depends on the strain because some strains flocculate more compactly than others, etc. The bottom line is that the only real way to estimate counts are with a microscope. But in the majority of homebrew situations, this is totally overkill, just infer your counts by monitoring everything else you can (lag time, attenuation, flavor formations, off flavors, etc.) then use those other data points to adjust your pitching rate.
I typically do what I think Denny is suggesting, tell the calculator you are repitching from slurry, and go very pessimistic about the non-yeast percentage and the thick-thin slurry numbers are.