Yes the choice of primary is somewhat random, so it's likely it will switch/connect to different ones at different times.Could it be they are both being connected all the time, but sometimes the router has one as the primary (is that the right word) and other times it has the other? And SINR is calculated from the primary frequency?
Very odd they both get the same speed, are 4G+ but report wildly different SINR (for obvious reasons). You'd expect if it was the actual SINR then the speed would differ quite a bit. Maybe that's a question for HUAWEI and how the B818 behaves on 4G+/LTE-A?
Yes I believe the SINR is reported only for the primary band. I don't believe the Huawei API gives any further insight into secondary band metrics (whereas something like NSG for Android does).
RF metrics being 'low' don't necessarily lead to poorer speeds - they are a measure of the radio medium and are not an indication of throughput speeds. However, saying that if all metrics are low then that could be the thing that is the limiting factor in achieving higher throughput.