v0.7.2: CRAN bugfix release

This version fixes a number of bugs and some regressions introduced in v0.7.1. It is available on CRAN.
As always, a huge thanks to everybody who took the time to file these issues.

What's New

• In position_quasirandom, a new orientation argument allows you to manually specify which axis is the 'grouped' or categorical axis. Passing orientation="x" is equivalent to the now-superseded argument groupOnX=TRUE . The parameter also accepts a value of "y" or NULL; the latter causes position_quasirandom to guess which axis is grouped (see below). This addition brings back the ability to change the orientation of the swarms without having to use coord_flip() (#82).
• position_quasirandom more intelligently guesses which axis is the grouped axis by default, and outputs a message to the user when the guess results in the axes being 'flipped', in case the guess is incorrect.
• position_beeswarm now warns the user once per session when using methods that discretize the data axis, and thus may shift the points unexpectedly (specifically, methods center, hex and square) (#77).

Bugfixes

• Fixed an issue in position_quasirandom where the user needed to specify e.g. aes(group = 1) or dodge.width=NULL (#83, #80)
• Fixed an issue in position_quasirandom where NAs in non-data axis aesthetics (e.g., shape) made the points disappear (#84)
• Fixed an issue in position_beeswarm where an error would occur if all the values of y were the same (#85)
• Fixed an issue in position_quasirandom where certain aesthetics would result in one group per observation (e.g., using aes(text=...) for ggplotly) and thus prevent any position adjustment (#78)