I know it's not a perfect comparison, but GA-6 and GA-7 were both 60R-37D (-ish) districts in 2012 as well. Not necessarily guaranteed to be safe.
TX-10 and TX-21 (Austin crack districts) each gave Romney around 60% but only barely went to Trump last year.
So just to repeat stop looking at the percentage change in the vote Trump won each district that takes in Nashville by more votes in 2020 than he did in 2016. That's the trend to look at not how the percent changed. 2020 had a lot higher turnout than 2016 so these percentage changes are skewed and look worse than they are. If the best the dems can do is turnout more voters and lose by more than these should be safe.
Without knowing how much higher turnout was in 2020 that's kinda a useless metric though. If turnout is brought high enough pretty much any raw margin will increase, unless the district is trending so fast it's bound to be gone anyway.
Besides that doesn't even tell the whole picture, once turnout is down again in the next cycle, does the raw margin stay the same or does the percentage persist with the lowered margins?