NetApp guys can correct me here, but I think it's not just 9.5, even in prior versions of OCUMs, LIFs do/did not show IOPS & Latency, only in cases where LIF is used for iSCSI data where you expect block IOPS.
Rest all LIFs will only show the MB/sec b'cos that's a true measure of how much data (Throughput) is received/sent through that LIF per-second. Therefore it says N/A (Not Applicable).
Qos Command (qos statistics latency show) provides Latency observed per I/O operation in the Network subsystem (Network), but it's overall and a good pointer towards what layer is contributing towards the total avg latency.
You are correct, QoS is good when you are investigating something real-time and there is a fire-fighting like situation. But, I also make use of OCUM to look at historic data (Node-Utilizaiton, /Aggr/Vol/Lun IOPS/Latency/MBps, except LIFs) and that helps me in understanding the trigger point.
Logic I can think of is : Rest all the LIFs which serves (CIFS/NFS) are dealing at file-evel so aren't aware, whereas the LIFs (end-point) that serves iSCSI [Block IOs] is aware.