There is a bit of a caveat in the counter performance, and this may be what you are seeing. The quad direction changes have a small window of software processing, and if another direct change occurs before that processing, we can miss stuff. It has only been a problem in very rare cases, usually involving simulated encoder inputs that have a high degree of jitter. If the servo is sitting almost still and is walking a few counts...logically, if not physically...up and down, and the servo's potential freq output is high, it may be causing back to back direction changes that we are missing.
I don't remember the exact number off the top of my head, but in general, we don't want to see a 2nd direction change any quicker than about 50 us.