After some thinking, I've got a theory: Deaths lag behind the infections by say 4 weeks. If the infections are falling after a peak, deaths are still rising or plateauing for 4 weeks. That means the ratio deaths/infections will produce a peak that lags the infection peak by 4 weeks, because the divisor is getting smaller while the dividend is still rising. Now if you look at the numbers for Israel, there is a plateau in the CFR for the last few weeks, it looks quite flat. However, e.g. Germany or Austria do have a pronounced peak there. The UK slightly less so.
So I would argue that we do see the CFR being lower than the peak we would expect without working vaccinations.
If I take a look on (rolling average) of cases and deaths, peaks are more or less at the same point (about 1 week for this peak and 2 weeks for the previous one, definitely not 4 weeks apart. In particular, now b)oth number of cases and deaths are decreasing.
So I would argue that we do see the CFR being lower than the peak we would expect without working vaccinations.