1.135 adds a new layer of polishing to source document sanitization. Which is to say, I've added some regular expression passes to the book processing that hunts down oddities like punctuation for italic text that's sitting outside italic tags (which can produce odd extraneous spaces, due to the text justification pulling apart anything that isn't in the same tag), or unnecessary empty paragraph tags left over from particularly troublesome source files.
This does mean that book the time it takes to process a book has probably gone up by a second or two, depending on the speed of your machine and the drive you're reading from, but the end result is a more polished product that should require less manual tweaking.
That is, assuming I haven't broken it entirely.