pick_main_stem changes order of tags, and does so differently for each census

[hmullerlandau]

pick_main_stem does not preserve the order of tags within a dataset, and it does not produce a consistent new order for different censuses. Thus, if pick_main_stem is applied to two different stem datasets of the same plot, the resulting "full" census datasets will have rows in different order by tag number. This generates problems if the resulting full data sets are then used with functions that assume rows are in the exact same order in different datasets, as for example assumed by all functions in the old CTFS R package. This means that, for example, calculating species-specific mortality rates using mortality.spp or for any other subset of the data will result in incorrect results. No error or warning messages are generated, however, so this problem could easily be overlooked. This problem was detected when applying pick_main_stem to the first two stem datasets for the Mpala plot.