The sample is only 22% Black? Really?
Just for reference, the 2015 primary was around 34% black, the 2016 presidential primary was 29% black and the 2018 general was 32% black. I think an average of those 3 would be a reasonable expectation for overall turnout (31.7%).
I'm not sure of the exact figures for the 2015 general, but it's hard to see Robert Gray of all people winning nearly 20% of the white vote (which is what would be mathematically required with a 22% black electorate going 9:1 in his favor).
Of course, for Hood to get 40% in a poll with a 22% black sample, the white share of the vote may be out of wack, too (assuming 85% of blacks in the poll support Hood, that'd require 27% of the non-black vote for Hood as well). Granted, Hood needs around 30% of the white vote to have a shot, so who knows.