Except if you watch the video I realign the HMD again after the INS is set to IFA to the same result. Regardless, I do not anticipate this is expected behavior.
I do realize the update frequency of SA data. However, in DCS, when it updates it should be accurately placed at the update interval, which you can clearly see from the video it is not. Designating an A/G target would produce the same result. I cannot reliably reproduce this bug - it seems to happen randomly but I haven't yet dedicated time to try to reproduce it. If you compare this video to a normal HMD alignment, the same "jump" occurs, but with a correctly functioning alignment it jumps down and then back to the original location. When this bug occurs, it does not first jump down and simply jumps up and out of alignment (but then can't be corrected).