In scanpy, `mean_var` used `@numba.jit(parallel=True)`, we don’t We should either port scanpy’s smart decorator or just pass `parallel=True` and let the users with many cores deal with it.
In scanpy,
mean_varused@numba.jit(parallel=True), we don’tWe should either port scanpy’s smart decorator or just pass
parallel=Trueand let the users with many cores deal with it.