Your problem is you don't want to listen to any of my arguments. And I'm right simply because of one simple fact: the risk of changing the source code cannot be zero, despite you arguing that it is. ...
The risk of inserting exponential transaction growth as Gavin wants to try to sneak in as a hard fork is, to me, greater than the risk of increasing the 21 million currency base figure.
Adding a feature as a soft fork to support sidechains is, from a technical structural standpoint, not necessarily a very risky thing. If it is implemented in a reasonably manner I'll definitely prefer it as the mainline Bitcoin codebase.
In some respects I actually would like to see sidechains implemented exercising a federated paradigm (instead of or in addition to a soft-fork) because development of such a solution has potential to mitigate against other forms of superior resource based attacks. And thus make them less likely to be attempted.