Seems similar behavior was reported in #610332, but the fix for that bug adds checks to see if aiccu is enabled and running instead of removing the sleep.d script entirely, as Jeroen suggests (I believe removing the script is the correct action as well).
#610332 is of interest because the fix was a non-maintainer upload, so I suspect the aiccu package's maintainer is no longer active. I'll see what I can find out.
