Edit: BTW, great article. Thank you.
2) Clients are buying your availability, they aren't buying bump-what-you're-doing-now. (That's a separate service, and ridiculously expensive.) If they find makework for you on the 29th, you tell them "Gotcha, I'll tell you when it is done and we'll count it against this month's allocation." If they balk at that, you can tell them that happily your email address works all 30 days of the month and that they can get in their requests quicker for faster turn-around.
Of course, the ultimate protection against clients being impossible to work with is not working with them, but -- again -- I'd wait until you actually had a problem before rejecting them for that.
The benefit for the client is that they don't need to review and sign a brand new Statement of Work whenever they want to change a button's color or tweak a feature.
Should your client need work that will seriously disrupt your schedule / happiness... I wrote my thoughts on that a few months back: http://planscope.io/blog/should-you-charge-more-for-a-rush-j...
1) In my experience - it doesn't happen.
2) The way I set it up to alleviate that fear, before I saw that it didn't happen, was to separate the booking time from the delivery time. So $client gets N days each month when they can ask for a booking, which will then get scheduled in over the next 30 days from the point they book it (so if the client requests N days on April 15th, they get booked in any time from Apr 15-Jun 14.)
This will also keep you from being surprised when someone wants their 10h/(month|week|whatever), and may actually prompt you to recommend how to use that time. They will pay you for leads on additional work because they see that it's a winning situation for them.