In any/all versions of PT, adding dither as the last plugin is only required if the session is 24 bit, but you are bouncing to 16 bit(as you might for creating a file to burn onto an audio CD).
If your session is 24 bit and want to EXPORT a file to 16 bit, the PT adds dither automatically. This is handy if you(as many do) record your final mix to a stereo track in the session(which will be at the same sample rate and bit depth of the session) and then export(not the same as Bounce to Disk, but works fine for those that prefer this method).
If you are sending your mixes out to be mastered, keep them at the session's sample rate and bit depth and let the mastering engineer take care of it.
My preference is to work at 24 bit/48K. If I am "self-mastering" a single song, then dither is among the processes I use on the master track(always the last process). BUT, if I am doing a collection of several songs, I will bounce all at 24/48>Import the finals into a new "mastering" session and process each song(so I can compare and make sure levels and tonality match). Then I use Track Commit to make the final masters(without dither) and Export Clips(and let PT do its automatic dithering)
