The SemEval limits are different from the ACL ones. SemEval strongly recommends that you stick to 4 pages if at all possible.
There were 148 notes in the test set. I’ll put all the statistics about the dataset in the task description paper, so for your system description papers, you don’t need to repeat that, you can just cite the task description paper. (SemEval will have a procedure this year that posts the BibTeX for all papers before the final proceedings are produced so that we can all easily cite each other.) Of course, if you need those stats to make a point in your system description paper, of course go ahead and do that.
As a side note, the reviewing procedure for SemEval this year will give you all a chance to look at the task description paper and send any feedback on it before it is finalized for the proceedings.
The SemEval website has been updated to include a timezone, GMT -12:00. As long as it’s still Mon 27 Feb 2017 anywhere on earth, you can still make your submission.