About the link you posted the gragh is not showing LiFePO4 chemistry.
I have an Elite Power Solutions system which consite of 20 GBS cells and uses a per cell sense boards for voltage and temperature, top balancing per cell shunts, a system shunt for current monitoring and coulomb counting (EMS) and 2 high current protective 400a relays.


The EMS CPU is programmed with these alarms and warnings are displayed at the bottom of the display screen after the word “Pack:”. There are seven alarms:
- Over voltage (highest cell is over 3.8V after a 3 second delay).
- Under voltage (lowest cell is below 2.8V after a 30 second delay).
- Over current (current exceeds 10C for 10 seconds).
- Over temperature (highest cell exceeds 150°F or 65°C).
- Under temperature (lowest cell is below 32°F or 0°C, charging is not allowed).
- Ground fault (There is a high voltage leakage greater than 2mA to the chassis pin).
- Unmanaged cells (The programmed number of cells does not equal the number of cells read).
SOC:
The EMS system has a minimum recorded resolution of +/- 0.5A so any current below that level is not accounted for. The system could be setup for a finer resolution, but the shunt value would have to be reduced which reduces the systems maximum current capability.
The system will reset the capacity to 100% whenever the pack voltage reaches 3.49 to 3.52 vpc on average, the pack current must be negative (charging) and typically under 35A and then it will set the capacity to 100%. These are all firmware variables.
With these things in place I don't really worry about the batteries. It's been almost 4 years now with everyday use and solar being the main source for charging 99% of the time using the settings I mentioned. Daily sundown to sunup usage 175-225ah.
If I didn't fulltime I would break the cells down for a capacity test after 4 years now but not possible seeing they are my power source and I boondock all the time but some day maybe if I come off the road.