Wasn't even that close in the end. 10 points is a pretty signifciant win in a race that deemed so close a week ago.
Generally in a referendum like this undecideds go to the stats quo in the voting booth.
your making that up
No. Status quo bias is an actual thing. Voting yes in such a referendum seems a lot more final than voting No. There may be future independence referendums. If Scotland voted yes then the decision would be made and most probably never revisited. Undecided voters therefore may vote "No" as more of a delaying tactic - they feel they can always come back and vote yes later. Humans have been shown to not like having their options reduced. This is why playing hard to get in relationships works. If someone feels they can have you whenever they want then why would they have you now? They would rather try with someone else so they have as many options as possible. Voting yes gives no future options; voting no gives the option to vote yes in 20 years.
If you look at polls for the last week or so a lot of them were saying 45% Yes, 50% No and 5% undecided. The last poll which gave the No vote a greater than 10% winning margin was published on 15 August. All subsequent polls predicted a closer final margin. They were wrong not because of any failure of methodology but because those who are undecided going into the polling booth are more likely to vote No.
The same effect was seen in the 1995 Quebec Referendum when all polls in the last week predicted an independence vote but they rejected it by a slender margin.