How to use the numbagg.decorators.ndreduce function in numbagg

@ndreduce([float32(float32), float64(float64)])
def nanstd(a):
    # for now, fix ddof=0
    ddof = 0
    asum = 0
    count = 0
    for ai in a.flat:
        if not np.isnan(ai):
            asum += ai
            count += 1
    if count > ddof:
        amean = asum / count
        asum = 0
        for ai in a.flat:
            if not np.isnan(ai):
                ai -= amean
                asum += ai * ai

@ndreduce([int64(int32), int64(int64), float32(float32), float64(float64)])
def nanmin(a):
    if not a.size:
        raise ValueError(
            "zero-size array to reduction operation fmin which has no identity"
        )
    amin = np.infty
    all_missing = 1
    for ai in a.flat:
        if ai <= amin:
            amin = ai
            all_missing = 0
    if all_missing:
        amin = np.nan
    return amin

@ndreduce([float32(float32), float64(float64)])
def nanvar(a):
    # for now, fix ddof=0
    ddof = 0
    asum = 0
    count = 0
    for ai in a.flat:
        if not np.isnan(ai):
            asum += ai
            count += 1
    if count > ddof:
        amean = asum / count
        asum = 0
        for ai in a.flat:
            if not np.isnan(ai):
                ai -= amean
                asum += ai * ai

@ndreduce([int64(int32), int64(int64), int64(float32), int64(float64)])
def count(a):
    non_missing = 0
    for ai in a.flat:
        if not np.isnan(ai):
            non_missing += 1
    return non_missing

@ndreduce([int64(int32), int64(int64), float32(float32), float64(float64)])
def nanmax(a):
    if not a.size:
        raise ValueError(
            "zero-size array to reduction operation fmax which has no identity"
        )
    amax = -np.infty
    all_missing = 1
    for ai in a.flat:
        if ai >= amax:
            amax = ai
            all_missing = 0
    if all_missing:
        amax = np.nan
    return amax

@ndreduce([int32(int32), int64(int64), float32(float32), float64(float64)])
def nansum(a):
    asum = 0
    for ai in a.flat:
        if not np.isnan(ai):
            asum += ai
    return asum

@ndreduce([int64(int32), int64(int64), int64(float32), int64(float64)])
def nanargmax(a):
    if not a.size:
        raise ValueError("attempt to get argmax of an empty sequence")
    amax = -np.infty
    idx = -1
    for i, ai in enumerate(a.flat):
        if ai > amax or (idx == -1 and not np.isnan(ai)):
            amax = ai
            idx = i
    if idx == -1:
        raise ValueError("All-NaN slice encountered")
    return idx

@ndreduce([float32(float32), float64(float64)])
def nanmean(a):
    asum = 0.0
    count = 0
    for ai in a.flat:
        if not np.isnan(ai):
            asum += ai
            count += 1
    if count > 0:
        return asum / count
    else:
        return np.nan

@ndreduce([int64(int32), int64(int64), int64(float32), int64(float64)])
def nanargmin(a):
    if not a.size:
        raise ValueError("attempt to get argmin of an empty sequence")
    amin = np.infty
    idx = -1
    for i, ai in enumerate(a.flat):
        if ai < amin or (idx == -1 and not np.isnan(ai)):
            amin = ai
            idx = i
    if idx == -1:
        raise ValueError("All-NaN slice encountered")
    return idx

@ndreduce([bool_(int32), bool_(int64), bool_(float32), bool_(float64)])
def anynan(a):
    f = False
    for ai in a.flat:
        if np.isnan(ai):
            f = True
            break
    return f