benchmark: consider the use of GOMAXPROCS in place of numCPU

[weikengchen]

This is related to benchmarking. When benchmarking in Golang, one can set an additional parameter to specify the number of CPUs allowed to that program, with a flag -cpu=2,4,8.
https://dev.to/mcaci/introduction-to-benchmarks-in-go-3cii

This parameter, interestingly, only affects runtime.GOMAXPROCS(0) but not runtime.numCPU(). Therefore, when one uses the flag above to limit the number of CPU cores when benchmarking, a number of functions in the codebase, such as FFT, will mispredict the number of CPUs available to it.