On what basis do you make this claim? Is it your position that the chip (LM2596) cannot be used in a converter with greater than 75% efficiency? I think it has more to do with the other components than the chip itself.

https://www.ti.com/lit/ds/symlink/lm2596.pdf Figure 7-3.

 

Doesn't that mean that it partially discharges some cells whilst it is waiting for the others to get fully charged? Won't it extend the charge time?
Any "ballasting" would reduce the efficiency of the charger.
I've recently used this one:
https://www.analog.com/en/products/ltc3639.html#product-overview
which will work up to 150V and has 12uA quiescent current.

expensive but good solution. Thanks))

 

https://www.ti.com/lit/ds/symlink/lm2596.pdf Figure 7-3.

It is the components, not the chip. You can do better when you select better components.
Since the LM2596 has a fairly low fixed switching frequency, you might try a different one with a higher fixed frequency or one with an adjustable switching frequency,

 

expensive but good solution. Thanks))

Exactly what I thought. However, it worked first time, and I didn't manage to blow up a single one during prototyping!