Some additional stats that might help understand the implications are that the not only is the jump in "respect" even larger among non-whites than whites, but its much larger among Dems than Republicans, among younger people than older people, and among those in rural areas than urban areas. That's kind of complicated, because the rural increase seems almost at odds with younger and non-white increases. Note that older people, whites, and conservatives continue to report the most respect for police, but the 1 year jump was greatest among younger people, liberals, non-whites, and rural dwellers.
The "jump" largely fake, by which I mean people are not reporting their true feelings about the police, but rather are giving strategic responses to send political messages.
The complex pattern of the jump suggests that different groups are trying to send different messages. Some conservatives and others who disagree with the extreme anti-police rhetoric of BLM are being reactionary and inflating their reported "respect" for the police as a way to express their disagreement with that rhetoric. That doesn't just mean white people, because the results show that both whites and non-whites increased in "respect" by over 10% points. But its quite likely that older non-whites also disagree with BLM tactics and rhetoric, which is mostly coming from millennials, and other surveys show that Hispanics are not that supportive of BLM.
At the same time, there could be millennial leftists and minorities who actually do not have much respect for the police but say they do in the survey to counter the idea that the clashes with police are partly due to a general lack of respect for the police and the law that the cops encounter.