MDS misalignment

[infotroph]

The MDS plots produced by the current version of plot_melt_distance.R seem to have scaling / alignment issues -- sample centroids aren't plotted near the points that come from those samples, and vector length for the extraction effect doesn't seem to line up well with estimates from adonis. This may be a graphics issue more than a computation issue -- when I run similar code in an interactive session, the resulting plots look OK.