I think the issue here is that the Trimetric does indeed show net amps across its shunt, but there is an alternative way to look at whether all those amps are going through the battery to supply the loads or some just go across the battery pos terminal common connection and the shunt's neg common connection without going down inside the battery.
-Doing a test for what amps the converter can do on its path from DC panel to battery. Batteries are jammed full at 100% SOC
-Turn on kettle via inverter. Trimetric shows -90 amps draw. Turn on converter. Tri shows -40 amps. This shows that the converter is supplying 50a.
Of interest perhaps, is that the instructions for the solar controller say do not run loads from the load terminals of the controller without the battery connected to the controller. Why not, if the battery cannot charge and discharge at the same time?
Perhaps what happens is that the battery "sees" different voltages as one net voltage wrt either loads or chargers and responds plus or minus to that net voltage diff with amps heading one way or the other?