Your IC is designed for 38 to 55V batteries with 35 to 55 Vinputs. My explanation is for Laptop chargers. The same philosophy may apply to Lithium secondary batteries but not Lead Acid which can tolerate 2.3V per 2V cell in continuous charge voltage ( with temp compensation for glass Matt types.) So the hysteresis may be as small as you like to regulate your voltage. 10mV for example as long as there is sufficient deadband to prevent shoot thru in the commutation between modes. This can vary depending on the delta V=I*ESR of the batteries which can be high enough to required performance testing of your intended batteries and thus cause hunting as noted by my colleague but with PWM on light duty might be small enough to be suppressed by a low ESR e-cap to mitigate ripple from this mode of using a load with the batteries while charger is connected. You may want to determine if battery current is needed to be monitored separately from charger current. In car or Golf Cart batteries using lead acid, this is not necessary. They often use 48~54V so CV is adequate but then you wont be running the golf cart with the charger connected (lol). But I digress, the datasheet is intended for Li-Ion cells so you may want to prevent the CV mode duration with a much larger hysteresis of 4.2 to 3.6V per cell while cutoff current is set to 10% of CC.
A battery charger does not like being above nominal fully charged cell voltage of 4.2V CV mode per cell for long periods of time and contributes to self-heating and aging more rapidly than not in use. the ideal cell voltage is 66% State of Charge (SoC) often used by some laptop drivers by Lenovo. The charge sequence is CC to 4.2V per cell and the transition from 3.8V to 4.2 CC then 4.2V CV mode until current drops below some pre-designed threshold like 5 to 10% of CC current levels. Thus when charger is connected all the time, the SoC may stay at 100% but only is actively boosting battery voltage when it drops from some predefined threshold of 100% and 99% (3.8~3.7V) then boosts the voltage to 4.2V for some minimal time till CV cutoff current is reach. This cycles more rapidly as the battery ages and causes more self-heat and aging and thus ages even more rapidly.
Now this arm-waving explanation is close enough to explain how battery chargers work while in use by sensing battery current independently of charger to load current. But it does not describe how a buck-boost DCDC converter works without a battery. In this case the hysteresis can be very small to reduce voltage error as there is no aging effects on capacitors with this tight voltage regulation.
That being said, DCDC converters for stability reasons do not use PID control but most just use PD control meaning Proportional and partial derivative to predict voltage changes by RC partial derivative across a R ratio feedback to regulator. This gives better regulation. But ultimately the load causes some voltage limited by the driver RdsOn and the loop gain on the error voltage. This full load voltage error is an measurement also of the driver resistance divided by the limited DC loop gain which is also limited for step load overshoot and stability reasons. Thus Load regulation errors of 1 to 2% reflect the output DC resistance of the power supply.
So to summarize, your graph shows a large hysteresis of perhaps 0.1V or more to minimize the cycle time and frequency of keeping the battery at 100% SoC. Ideally this should be kept as minimal as possible to reduce self-heating and battery aging.
But this has nothing to do with battery-less Buck-Boost DC-DC regulators.
In fact laptops are mostly standardized now with 19.5V chargers regardless of the battery array size and voltage. The motherboard on the other side does not need more than 5V to operate the peripherals. (Maybe 12V at one time) It is not know to me but it is possible that there is no boost regulator needed for the laptop battery chargers and they are always lower voltage like 2P3S 11.1V or 2P4S 14.8V (3.7V/cell). So it is always in Buck mode. Which makes things much cheaper to design and lower demand/ surge current.
However inspite of this Buck mode used some mobiles still have large input capacitance and this tends to pit the gold flash plating on IOS lightning connectors so I recommend if you want low live surge damage to your lightning USB cables to connect the USB end last . ( I have worn out 3 such cables from experience;)
The datasheet says Cutoff is 3 to 5 % of CC current which might give you an extra 5% of battery capacity in the short term but -25% in the long term if you reduced the capacity by only charging to 4.1V or 4.0V and get much more charge life cycles than typical and maybe even the 500 cycles some suggest instead of 350 full load cycles. goto Battery University website for latest info on Li-Ion longevity test results.