That document asserts that they identified 28,000 noncitizen registered voters, but the methodology seems to have been based on name matches where for 28K voters, they couldn't find positive proof that those names were of citizens. But that doesn't actually mean they aren't.
And then they extrapolate their sample to the rest of the country and conclude that there are over a quarter of a million noncitizens who have registered to vote nationwide.
This is just methodologically silly.