3 stages to be assembled into a 1-stage pipeline: 1. input unpacking & adjust for magnitude comparison (mostly abs) not using existing unpack stage since that computes unbiased exponent which is unnecessary and undesired. 2. compare inputs and nan-detection 3. mux output with optional nan-quieting https://bugs.libre-soc.org/show_bug.cgi?id=1135#c20 since the FPSCR classes task is quite small, I think we should allocate the budget for that task through this task.