How to use the yasa.others.trimbothstd function in yasa

assert hypno.size == data.size, 'Hypno must have same size as data.'
        unique_hypno = np.unique(hypno)
        logger.info('Number of unique values in hypno = %i', unique_hypno.size)
        # Check include
        assert include is not None, 'include cannot be None if hypno is given'
        include = np.atleast_1d(np.asarray(include))
        assert include.size >= 1, '`include` must have at least one element.'
        assert hypno.dtype.kind == include.dtype.kind, ('hypno and include '
                                                        'must have same dtype')
        if not np.in1d(hypno, include).any():
            logger.error('None of the stages specified in `include` '
                         'are present in hypno. Returning None.')
            return None

    # Check data amplitude
    data_trimstd = trimbothstd(data, cut=0.10)
    data_ptp = np.ptp(data)
    logger.info('Number of samples in data = %i', data.size)
    logger.info('Sampling frequency = %.2f Hz', sf)
    logger.info('Data duration = %.2f seconds', data.size / sf)
    logger.info('Trimmed standard deviation of data = %.4f uV', data_trimstd)
    logger.info('Peak-to-peak amplitude of data = %.4f uV', data_ptp)
    if not(1 < data_trimstd < 1e3 or 1 < data_ptp < 1e6):
        logger.error('Wrong data amplitude. Unit must be uV. Returning None.')
        return None

    if 'rel_pow' not in thresh.keys():
        thresh['rel_pow'] = 0.20
    if 'corr' not in thresh.keys():
        thresh['corr'] = 0.65
    if 'rms' not in thresh.keys():
        thresh['rms'] = 1.5

# Hilbert power (to define the instantaneous frequency / power)
    n = data_sigma.size
    nfast = next_fast_len(n)
    analytic = signal.hilbert(data_sigma, N=nfast)[:n]
    inst_phase = np.angle(analytic)
    inst_pow = np.square(np.abs(analytic))
    # inst_freq = sf / 2pi * 1st-derivative of the phase of the analytic signal
    inst_freq = (sf / (2 * np.pi) * np.ediff1d(inst_phase))

    # Let's define the thresholds
    if hypno is None:
        thresh_rms = mrms.mean() + thresh['rms'] * trimbothstd(mrms, cut=0.10)
    else:
        thresh_rms = mrms[mask].mean() + thresh['rms'] * \
            trimbothstd(mrms[mask], cut=0.10)

    # Avoid too high threshold caused by Artefacts / Motion during Wake.
    thresh_rms = min(thresh_rms, 10)
    idx_rel_pow = (rel_pow >= thresh['rel_pow']).astype(int)
    idx_mcorr = (mcorr >= thresh['corr']).astype(int)
    idx_mrms = (mrms >= thresh_rms).astype(int)
    idx_sum = (idx_rel_pow + idx_mcorr + idx_mrms).astype(int)

    # Make sure that we do not detect spindles in REM or Wake if hypno != None
    if hypno is not None:
        idx_sum[~mask] = 0

    # For debugging
    logger.info('Moving RMS threshold = %.3f', thresh_rms)
    logger.info('Number of supra-theshold samples for relative power = %i',
                idx_rel_pow.sum())

logger.info('Number of unique values in hypno = %i', unique_hypno.size)
        # Check include
        assert include is not None, 'include cannot be None if hypno is given'
        include = np.atleast_1d(np.asarray(include))
        assert include.size >= 1, '`include` must have at least one element.'
        assert hypno.dtype.kind == include.dtype.kind, ('hypno and include '
                                                        'must have same dtype')
        if not np.in1d(hypno, include).any():
            logger.error('None of the stages specified in `include` '
                         'are present in hypno. Returning None.')
            return None

    # Check data amplitude
    # times = np.arange(data.size) / sf
    data = np.vstack((loc, roc))
    loc_trimstd = trimbothstd(loc, cut=0.10)
    roc_trimstd = trimbothstd(roc, cut=0.10)
    loc_ptp, roc_ptp = np.ptp(loc), np.ptp(roc)
    logger.info('Number of samples in data = %i', data.shape[1])
    logger.info('Original sampling frequency = %.2f Hz', sf)
    logger.info('Data duration = %.2f seconds', data.shape[1] / sf)
    logger.info('Trimmed standard deviation of LOC = %.4f uV', loc_trimstd)
    logger.info('Trimmed standard deviation of ROC = %.4f uV', roc_trimstd)
    logger.info('Peak-to-peak amplitude of LOC = %.4f uV', loc_ptp)
    logger.info('Peak-to-peak amplitude of ROC = %.4f uV', roc_ptp)
    if not(1 < loc_trimstd < 1e3 or 1 < loc_ptp < 1e6):
        logger.error('Wrong LOC amplitude. Unit must be uV. Returning None.')
        return None
    if not(1 < roc_trimstd < 1e3 or 1 < roc_ptp < 1e6):
        logger.error('Wrong ROC amplitude. Unit must be uV. Returning None.')
        return None

method='corr', interp=True)
    _, mrms = moving_transform(data_sigma, data, sf, window=.3, step=.1,
                               method='rms', interp=True)

    # Hilbert power (to define the instantaneous frequency / power)
    n = data_sigma.size
    nfast = next_fast_len(n)
    analytic = signal.hilbert(data_sigma, N=nfast)[:n]
    inst_phase = np.angle(analytic)
    inst_pow = np.square(np.abs(analytic))
    # inst_freq = sf / 2pi * 1st-derivative of the phase of the analytic signal
    inst_freq = (sf / (2 * np.pi) * np.ediff1d(inst_phase))

    # Let's define the thresholds
    if hypno is None:
        thresh_rms = mrms.mean() + thresh['rms'] * trimbothstd(mrms, cut=0.10)
    else:
        thresh_rms = mrms[mask].mean() + thresh['rms'] * \
            trimbothstd(mrms[mask], cut=0.10)

    # Avoid too high threshold caused by Artefacts / Motion during Wake.
    thresh_rms = min(thresh_rms, 10)
    idx_rel_pow = (rel_pow >= thresh['rel_pow']).astype(int)
    idx_mcorr = (mcorr >= thresh['corr']).astype(int)
    idx_mrms = (mrms >= thresh_rms).astype(int)
    idx_sum = (idx_rel_pow + idx_mcorr + idx_mrms).astype(int)

    # Make sure that we do not detect spindles in REM or Wake if hypno != None
    if hypno is not None:
        idx_sum[~mask] = 0

    # For debugging

assert hypno.size == data.size, 'Hypno must have same size as data.'
        unique_hypno = np.unique(hypno)
        logger.info('Number of unique values in hypno = %i', unique_hypno.size)
        # Check include
        assert include is not None, 'include cannot be None if hypno is given'
        include = np.atleast_1d(np.asarray(include))
        assert include.size >= 1, '`include` must have at least one element.'
        assert hypno.dtype.kind == include.dtype.kind, ('hypno and include '
                                                        'must have same dtype')
        if not np.in1d(hypno, include).any():
            logger.error('None of the stages specified in `include` '
                         'are present in hypno. Returning None.')
            return None

    # Check data amplitude
    data_trimstd = trimbothstd(data, cut=0.10)
    data_ptp = np.ptp(data)
    logger.info('Number of samples in data = %i', data.size)
    logger.info('Sampling frequency = %.2f Hz', sf)
    logger.info('Data duration = %.2f seconds', data.size / sf)
    logger.info('Trimmed standard deviation of data = %.4f uV', data_trimstd)
    logger.info('Peak-to-peak amplitude of data = %.4f uV', data_ptp)
    if not(1 < data_trimstd < 1e3 or 1 < data_ptp < 1e6):
        logger.error('Wrong data amplitude. Unit must be uV. Returning None.')
        return None

    # Check if we can downsample to 100 or 128 Hz
    if downsample is True and sf > 128:
        if sf % 100 == 0 or sf % 128 == 0:
            new_sf = 100 if sf % 100 == 0 else 128
            fac = int(sf / new_sf)
            sf = new_sf

# Check include
        assert include is not None, 'include cannot be None if hypno is given'
        include = np.atleast_1d(np.asarray(include))
        assert include.size >= 1, '`include` must have at least one element.'
        assert hypno.dtype.kind == include.dtype.kind, ('hypno and include '
                                                        'must have same dtype')
        if not np.in1d(hypno, include).any():
            logger.error('None of the stages specified in `include` '
                         'are present in hypno. Returning None.')
            return None

    # Check data amplitude
    # times = np.arange(data.size) / sf
    data = np.vstack((loc, roc))
    loc_trimstd = trimbothstd(loc, cut=0.10)
    roc_trimstd = trimbothstd(roc, cut=0.10)
    loc_ptp, roc_ptp = np.ptp(loc), np.ptp(roc)
    logger.info('Number of samples in data = %i', data.shape[1])
    logger.info('Original sampling frequency = %.2f Hz', sf)
    logger.info('Data duration = %.2f seconds', data.shape[1] / sf)
    logger.info('Trimmed standard deviation of LOC = %.4f uV', loc_trimstd)
    logger.info('Trimmed standard deviation of ROC = %.4f uV', roc_trimstd)
    logger.info('Peak-to-peak amplitude of LOC = %.4f uV', loc_ptp)
    logger.info('Peak-to-peak amplitude of ROC = %.4f uV', roc_ptp)
    if not(1 < loc_trimstd < 1e3 or 1 < loc_ptp < 1e6):
        logger.error('Wrong LOC amplitude. Unit must be uV. Returning None.')
        return None
    if not(1 < roc_trimstd < 1e3 or 1 < roc_ptp < 1e6):
        logger.error('Wrong ROC amplitude. Unit must be uV. Returning None.')
        return None

    # Check if we can downsample to 100 or 128 Hz