lp_dur
can be computed using lena columns (utterances/utterances_length/vfxs etc.) or with the 'vcm_type' column.
Problem : vcm_type can be NA when derived from lena but the linguistic info exists in lena form. The metrics functions are first checking for the presence of vcm_type and use it, they should use lena's column in priority.
metrics: