Daily deaths from COVID-19 are not when those deaths happened, but when they were reported from state and local sources. Many of reported deaths happened earlier, even 2-3 weeks earlier. Many states now report deaths that are thought to have been probably caused by COVID-19, not just tested cases. And these probable deaths are often reported in batches making daily figures appear to spike.
The "Provisional" CDC weekly figures shown here are more precise and detailed than so-called daily reports (e.g., they separate deaths that involved pneumonia or flu). But weekly death figures for recent weeks are incomplete ("provisional"), as the CDC waits for more death certificates.
After reaching a low of 3,519 deaths for the week ending June 27, COVID-19 deaths rose to 4,362 in the week ending July 11. Those COVID-19 deaths accounted for 8.4% of total deaths from all causes. The most recent two weeks look much lower than 4,362, but might end up higher after delayed death certificates are added.
In short, it does appear from the CDC count that COVID-19 deaths increased in July. But it was hardly the "surge" that many news reports suggested. The most recent credible weekly figure so far (4362) was little more than a fourth of the April peak of 16,982.
New cases have declined for 2-4 weeks in states where they had been rising such as Arizona, Texas and to a lesser extent Florida and Southern California. Due to the time lag between infection and deaths should begin to fall soon in Arizona and Texas, followed by Florida and California. I do not expect the nationwide CDC weekly figures to increase in August. Deaths have probably peaked. It wasn't a "second wave" but the first wave moving into new areas with the least "herd immunity."