Hi Thomas,
(1.) Well, do whatever you wish. As a long-time tech writer and GUI designer, though, I can tell you that, statistically, when users must "dig around" (consult documentation or fuss with settings) for basic functionality, such as moving the diagram in a diagramming program, they're much more likely to keep looking for an equivalent program that's more intuitively designed.
A standard Pan tool (button graphic examples here:
https://www.google.ca/search?q=drag+icon&source=lnms&tbm=isch) is something any user would recognize at first sight. It wouldn't interfere with your existing functionality; it'd just be a way to effectively tell a new user, "Want to pan your diagram? Yes, there's a way to do it—hunt around and you'll find it!"
All programs have keyboard shortcuts (in this case, holding the R mouse button), and committed users who like to work that way take the trouble to learn them. That's how it's always been. But you're not doing anyone a favor by assuming _everyone_ is such a user. You're just screening out people without the time, patience and/or attention skills to learn non-intuitive actions.
(2.) Yes, you're right: Double-clicking an existing label makes it editable. If anything, though, that _strengthens_ the argument for the 2nd click to create a label on an object that doesn't have one, too. In other words: Double-click = label mode, and the user needs remember only one action instead of two. Believe me, simpler is better. And again, I'm not suggesting you replace any existing functionality; this is just adding _extra_, intuitive functionality for those who prefer to work that way. What could possibly be wrong with that (other than possibly not having to be so rigid, LOL)?