Travis came up with a bug from the new ggplot2 v3 (Tokyo Drift): https://travis-ci.org/grunwaldlab/poppr/jobs/400114394#L1575-L1577

  ── 1. Failure: mlg.table produces barplots (@test-plots.R#80)  ─────────────────
  pt$mapping$x not identical to ptb$mapping$x.
  Objects equal but not identical

This was caused by my attempt to test plots before vdiffr was a thing, but I think this may be the better way to go (and it may make things go faster in the long run).