The nature of these lithium based batteries do not provide for a very consistent SOC vs voltage curve like with lead acid batteries. I have seen some of my batteries report a low SOC while the voltage still looks "high", just like your screen shots.
While I have not worked specifically with these wall mount units, I would think the BMS would be very similar to the LL units. When using multiple batteries in a cluster it is good to give the cluster a full cycle every now and then; what I mean by that is take it down to 20'ish% SOC then charge all the way up to 100% SOC in the same day if possible. This will help to get the BMS of each battery back into sync. I have my system setup/scripted to do this every 30 days. I have tried going longer but the SOC drift/variation from battery to battery seems to get even more out of sync and starts to throw off the overall SOC calculations (36 batteries/BMSs involved with this particular setup).
Even with the BMS SOC variation from battery to battery, I would still trust the reported SOC over the reported volts as a basis for actual SOC.
Here is the charge/discharge curve from one of the 48v-LL battery manuals. The voltage will "level out" through a very wide range of SOC then very quickly spike up or down depending on if you are charging or discharging. I believe the wall mount units are using a 280ah cell vs the 100ah in the LL units, but the battery/cell chemistry should be the same so the charge/discharge curves should be relevant.
Note how the voltage reports higher while charging, and lower while discharging. This would also throw off any voltage based SOC tracking.
J