Resampling of time when jitter time is turned on

When time marginalization is used, time is resampled by interpolating time posterior at its values at discrete time samples. However, I think the interpolation is not needed when jitter time is turned on. My understanding of time jitter is to model time by t_discrete + t_jitter, where t_discrete = (n + 1 / 2) * delta t and - delta_tc / 2 < t_jitter < delta_tc / 2. So I think a correct way to resample time is to sample t_discrete from discrete time samples with posterior weights and increment it by t_jitter. See the thread here.