You have a slight miscalculation due to mixing amps at 5V and amps at 3.6V(avg). Calculate for output power, divide by efficiency estimate and then work backwards to get the input current at average battery voltage. (It's good to calculate the input current at min & max cell voltages as well, for designing the wiring, fusing and DC/DC. For calculating Ah required, the average will do.)
80% is a poor efficiency target. Aim higher, at least 90% - say 93% for the DCDC and 97% for wires, fuses, etc - you'll save it back on cells, and thermal management becomes easier, so you save on heatsinking as well. After all, 20% of 8A*5V is whopping 8 watts you don't want to dissipate on a PCB!
The sane choices are either 1s with synchronous boost, or 2s with synchronous buck. Non-syncronous are way too inefficient for such high power, IMHO.
In a thermally coupled, small 2s system with quality brand cells, there is absolutely no need for center tap monitoring or balancing.
Remember to have a low-voltage cutoff, and make sure leakage after the disconnect doesn't kill your pack.
If you go for a boost, a separate FET switch to prevent unlimited current through the boost diode (or the body diode of the synchronous switch) is highly recommended. Integrated "e-fuse" power switch ICs do exist for this purpose.
Personally I would do with paralleled 18650's, but nothing fundamentally wrong with pouch cells, either. It just seems to me that still in 2019, it's easier to get good brand 18650's from gray brokers, than proper pouch cells. In li-ion cell safety, brands do matter.
Remember traditional (single-use) fusing right at the battery, in case any electronic protections fail, to prevent fires.
What are your requirements on the charging of power bank? This defines whether you can find an easy off-the-shelf charger IC or not.